1.Deformable ConvNets v2: More Deformable, Better Results pdf
The superior performance of Deformable Convolutional Networks arises from its ability to adapt to the geometric variations of objects. Through an examination of its adaptive behavior, we observe that while the spatial support for its neural features conforms more closely than regular ConvNets to object structure, this support may nevertheless extend well beyond the region of interest, causing features to be influenced by irrelevant image content. To address this problem, we present a reformulation of Deformable ConvNets that improves its ability to focus on pertinent image regions, through increased modeling power and stronger training. The modeling power is enhanced through a more comprehensive integration of deformable convolution within the network, and by introducing a modulation mechanism that expands the scope of deformation modeling. To effectively harness this enriched modeling capability, we guide network training via a proposed feature mimicking scheme that helps the network to learn features that reflect the object focus and classification power of R-CNN features. With the proposed contributions, this new version of Deformable ConvNets yields significant performance gains over the original model and produces leading results on the COCO benchmark for object detection and instance segmentation.
2.Integrated Object Detection and Tracking with Tracklet-Conditioned Detection pdf
Accurate detection and tracking of objects is vital for effective video understanding. In previous work, the two tasks have been combined in a way that tracking is based heavily on detection, but the detection benefits marginally from the tracking. To increase synergy, we propose to more tightly integrate the tasks by conditioning the object detection in the current frame on tracklets computed in prior frames. With this approach, the object detection results not only have high detection responses, but also improved coherence with the existing tracklets. This greater coherence leads to estimated object trajectories that are smoother and more stable than the jittered paths obtained without tracklet-conditioned detection. Over extensive experiments, this approach is shown to achieve state-of-the-art performance in terms of both detection and tracking accuracy, as well as noticeable improvements in tracking stability.
3.Label-Noise Robust Generative Adversarial Networks pdf
Generative adversarial networks (GANs) are a framework that learns a generative distribution through adversarial training. Recently, their class conditional extensions (e.g., conditional GAN (cGAN) and auxiliary classifier GAN (AC-GAN)) have attracted much attention owing to their ability to learn the disentangled representations and to improve the training stability. However, their training requires the availability of large-scale accurate class-labeled data, which are often laborious or impractical to collect in a real-world scenario. To remedy the drawback, we propose a novel family of GANs called label-noise robust GANs (rGANs), which, by incorporating a noise transition model, can learn a clean label conditional generative distribution even when training labels are noisy. In particular, we propose two variants: rAC-GAN, which is a bridging model between AC-GAN and the noise-robust classification model, and rcGAN, which is an extension of cGAN and is guaranteed to learn the clean label conditional distribution in an optimal condition. In addition to providing the theoretical background, we demonstrate the effectiveness of our models through extensive experiments using diverse GAN configurations, various noise settings, and multiple evaluation metrics (in which we tested 402 patterns in total).
4.Class-Distinct and Class-Mutual Image Generation with GANs pdf
We describe a new problem called class-distinct and class-mutual (DM) image generation. Typically in class-conditional image generation, it is assumed that there are no intersections between classes, and a generative model is optimized to fit discrete class labels. However, in real-world scenarios, it is often required to handle data in which class boundaries are ambiguous or unclear. For example, data crawled from the web tend to contain mislabeled data resulting from confusion. Given such data, our goal is to construct a generative model that can be controlled for class specificity, which we employ to selectively generate class-distinct and class-mutual images in a controllable manner. To achieve this, we propose novel families of generative adversarial networks (GANs) called class-mixture GAN (CMGAN) and class-posterior GAN (CPGAN). In these new networks, we redesign the generator prior and the objective function in auxiliary classifier GAN (AC-GAN), then extend these to class-mixture and arbitrary class-overlapping settings. In addition to an analysis from an information theory perspective, we empirically demonstrate the effectiveness of our proposed models for various class-overlapping settings (including synthetic to real-world settings) and tasks (i.e., image generation and image-to-image translation).
5.FineGAN: Unsupervised Hierarchical Disentanglement for Fine-Grained Object Generation and Discovery pdf
We propose FineGAN, a novel unsupervised GAN framework, which disentangles the background, object shape, and object appearance to hierarchically generate images of fine-grained object categories. To disentangle the factors without any supervision, our key idea is to use information theory to associate each factor to a latent code, and to condition the relationships between the codes in a specific way to induce the desired hierarchy. Through extensive experiments, we show that FineGAN achieves the desired disentanglement to generate realistic and diverse images belonging to fine-grained classes of birds, dogs, and cars. Using FineGAN's automatically learned features, we also cluster real images as a first attempt at solving the novel problem of unsupervised fine-grained object category discovery. Our video demo can be found at this https URL.
6.Understanding and Improving Kernel Local Descriptors pdf
We propose a multiple-kernel local-patch descriptor based on efficient match kernels from pixel gradients. It combines two parametrizations of gradient position and direction, each parametrization provides robustness to a different type of patch mis-registration: polar parametrization for noise in the patch dominant orientation detection, Cartesian for imprecise location of the feature point. Combined with whitening of the descriptor space, that is learned with or without supervision, the performance is significantly improved. We analyze the effect of the whitening on patch similarity and demonstrate its semantic meaning. Our unsupervised variant is the best performing descriptor constructed without the need of labeled data. Despite the simplicity of the proposed descriptor, it competes well with deep learning approaches on a number of different tasks.
7.See before you see: Real-time high speed motion prediction using fast aperture-robust event-driven visual flow pdf
Optical flow is a crucial component of the feature space for early visual processing of dynamic scenes especially in new applications such as self-driving vehicles, drones and autonomous robots. The dynamic vision sensors are well suited for such applications because of their asynchronous, sparse and temporally precise representation of the visual dynamics. Many algorithms proposed for computing visual flow for these sensors suffer from the aperture problem as the direction of the estimated flow is governed by the curvature of the object rather than the true motion direction. Some methods that do overcome this problem by temporal windowing under-utilize the true precise temporal nature of the dynamic sensors. In this paper, we propose a novel multi-scale plane fitting based visual flow algorithm that is robust to the aperture problem and also computationally fast and efficient. Our algorithm performs well in many scenarios ranging from fixed camera recording simple geometric shapes to real world scenarios such as camera mounted on a moving car and can successfully perform event-by-event motion estimation of objects in the scene to allow for predictions of upto 500 ms i.e. equivalent to 10 to 25 frames with traditional cameras.
8.Unprocessing Images for Learned Raw Denoising pdf
Machine learning techniques work best when the data used for training resembles the data used for evaluation. This holds true for learned single-image denoising algorithms, which are applied to real raw camera sensor readings but, due to practical constraints, are often trained on synthetic image data. Though it is understood that generalizing from synthetic to real data requires careful consideration of the noise properties of image sensors, the other aspects of a camera's image processing pipeline (gain, color correction, tone mapping, etc) are often overlooked, despite their significant effect on how raw measurements are transformed into finished images. To address this, we present a technique to "unprocess" images by inverting each step of an image processing pipeline, thereby allowing us to synthesize realistic raw sensor measurements from commonly available internet photos. We additionally model the relevant components of an image processing pipeline when evaluating our loss function, which allows training to be aware of all relevant photometric processing that will occur after denoising. By processing and unprocessing model outputs and training data in this way, we are able to train a simple convolutional neural network that has 14%-38% lower error rates and is 9x-18x faster than the previous state of the art on the Darmstadt Noise Dataset, and generalizes to sensors outside of that dataset as well.
9.Bilinear Parameterization For Differentiable Rank-Regularization pdf
Low rank approximation is a commonly occurring problem in many computer vision and machine learning applications. There are two common ways of optimizing the resulting models. Either the set of matrices with a given sought rank can be explicitly parametrized using a bilinear factorization, or low rank can be implicitly enforced using regularization terms penalizing non-zero singular values. While the former results in differentiable problems that can be efficiently optimized using local quadratic approximation the latter are typically not differentiable (sometimes even discontinuous) and require splitting methods such as Alternating Direction Method of Multipliers (ADMM). It is well known that while ADMM makes rapid improvements the first couple of iterations convergence to the exact minimizer can be tediously slow. On the other hand regularization formulations can in certain cases come with theoretical optimality guarantees.
In this paper we show how many non-differentiable regularization methods can be reformulated into smooth objectives using bilinear parameterization. This opens up the possibility of using second order methods such as Levenberg--Marquardt (LM) and Variable Projection (VarPro) to achieve accurate solutions for ill-conditioned problems. We show on several real and synthetic experiments that our second order formulation converges to substantially more accurate solutions than what ADMM formulations provide in a reasonable amount of time.
10.Automatic Face Aging in Videos via Deep Reinforcement Learning pdf
This paper presents a novel approach for synthesizing automatically age-progressed facial images in video sequences using Deep Reinforcement Learning. The proposed method models facial structures and the longitudinal face-aging process of given subjects coherently across video frames. The approach is optimized using a long-term reward, Reinforcement Learning function with deep feature extraction from Deep Convolutional Neural Network. Unlike previous age-progression methods that are only able to synthesize an aged likeness of a face from a single input image, the proposed approach is capable of age-progressing facial likenesses in videos with consistently synthesized facial features across frames. In addition, the deep reinforcement learning method guarantees preservation of the visual identity of input faces after age-progression. Results on videos of our new collected aging face AGFW-v2 database demonstrate the advantages of the proposed solution in terms of both quality of age-progressed faces, temporal smoothness, and cross-age face verification.
11.MobiFace: A Lightweight Deep Learning Face Recognition on Mobile Devices pdf
Deep neural networks have been widely used in numerous computer vision applications, particularly in face recognition. However, deploying deep neural network face recognition on mobile devices is still limited since most high-accuracy deep models are both time and GPU consumption in the inference stage. Therefore, developing a lightweight deep neural network is one of the most promising solutions to deploy face recognition on mobile devices. Such the lightweight deep neural network requires efficient memory with small number of weights representation and low cost operators. In this paper a novel deep neural network named MobiFace, which is simple but effective, is proposed for productively deploying face recognition on mobile devices. The experimental results have shown that our lightweight MobiFace is able to achieve high performance with 99.7% on LFW database and 91.3% on large-scale challenging Megaface database. It is also eventually competitive against large-scale deep-networks face recognition while significant reducing computational time and memory consumption.
12.Fast Object Detection in Compressed Video pdf
Object detection in videos has drawn increasing attention recently since it is more important in real scenarios. Most of the deep learning methods for video analysis use convolutional neural networks designed for image-wise parsing in a video stream. But they usually ignore the fact that a video is generally stored and transmitted in a compressed data format. In this paper, we propose a fast object detection model that incorporates light-weight motion-aided memory network (MMNet), which can be directly used for H.264 compressed video. MMNet has two major advantages: 1) For a group of successive pictures (GOP) in a compressed video stream, it runs the heavy computational network for I-frames, i.e. a few reference frames in videos, while a light-weight memory network is designed to generate features for prediction frames called P-frames; 2) Unlike establishing an additional network to explicitly model motion among frames, we directly take full advantage of both motion vectors and residual errors that are all encoded in a compressed video. Such signals maintain spatial variations and are freely available. To our best knowledge, the MMNet is the first work that explores a convolutional detector on a compressed video and a motion-based memory in order to achieve significant speedup. Our model is evaluated on the large-scale ImageNet VID dataset, and the results show that it is about 3x times faster than single image detector R-FCN and 10x times faster than high performance detectors like FGFA and MANet.
13.Understanding the Importance of Single Directions via Representative Substitution pdf
Understanding the internal representations of deep neural networks (DNNs) is crucial for explaining their behaviors. The interpretation of individual units which are neurons in MLP or convolution kernels in convolutional network have been payed much attentions since they act as a fundamental role. However, recent research (Morcos et al. 2018) presented a counterintuitive phenomenon which suggested an individual unit with high class selectivity, called interpretable units, had poor contributions to generalization of DNNs. In this work, we provide a new perspective to understand this counterintuitive phenomenon which we argue actually it makes sense when we introduce Representative Substitution (RS). Instead of individually selective units with classes, the RS refers to independence of an unit's representations in the same layer without any annotation. Our experiments demonstrate that interpretable units have low RS which are not important to the network's generalization. The RS provides new insights to the interpretation of DNNs and suggests that we need focus on the independence and relationship of the representations.
14.Dense xUnit Networks pdf
Deep net architectures have constantly evolved over the past few years, leading to significant advancements in a wide array of computer vision tasks. However, besides high accuracy, many applications also require a low computational load and limited memory footprint. To date, efficiency has typically been achieved either by architectural choices at the macro level (e.g. using skip connections or pruning techniques) or modifications at the level of the individual layers (e.g. using depth-wise convolutions or channel shuffle operations). Interestingly, much less attention has been devoted to the role of the activation functions in constructing efficient nets. Recently, Kligvasser et al. showed that incorporating spatial connections within the activation functions, enables a significant boost in performance in image restoration tasks, at any given budget of parameters. However, the effectiveness of their xUnit module has only been tested on simple small models, which are not characteristic of those used in high-level vision tasks. In this paper, we adopt and improve the xUnit activation, show how it can be incorporated into the DenseNet architecture, and illustrate its high effectiveness for classification and image restoration tasks alike. While the DenseNet architecture is extremely efficient to begin with, our dense xUnit net (DxNet) can typically achieve the same performance with far fewer parameters. For example, on ImageNet, our DxNet outperforms a ReLU-based DenseNet having 30% more parameters and achieves state-of-the-art results for this budget of parameters. Furthermore, in denoising and super-resolution, DxNet significantly improves upon all existing lightweight solutions, including the xUnit-based nets of Kligvasser et al.
15.Eliminating Exposure Bias and Loss-Evaluation Mismatch in Multiple Object Tracking pdf
Identity Switching remains one of the main difficulties Multiple Object Tracking (MOT) algorithms have to deal with. Many state-of-the-art approaches now use sequence models to solve this problem but their training can be affected by biases that decrease their efficiency. In this paper, we introduce a new training procedure that confronts the algorithm to its own mistakes while explicitly attempting to minimize the number of switches, which results in better training. We propose an iterative scheme of building a rich training set and using it to learn a scoring function that is an explicit proxy for the target tracking metric. Whether using only simple geometric features or more sophisticated ones that also take appearance into account, our approach outperforms the state-of-the-art on several MOT benchmarks.
16.GarNet: A Two-stream Network for Fast and Accurate 3D Cloth Draping pdf
While Physics-Based Simulation (PBS) can highly accurately drape a 3D garment model on a 3D body, it remains too costly for real-time applications, such as virtual try-on. By contrast, inference in a deep network, that is, a single forward pass, is typically quite fast. In this paper, we leverage this property and introduce a novel architecture to fit a 3D garment template to a 3D body model. Specifically, we build upon the recent progress in 3D point-cloud processing with deep networks to extract garment features at varying levels of detail, including point-wise, patch-wise and global features. We then fuse these features with those extracted in parallel from the 3D body, so as to model the cloth-body interactions. The resulting two-stream architecture is trained with a loss function inspired by physics-based modeling, and delivers realistic garment shapes whose 3D points are, on average, less than 1.5cm away from those of a PBS method, while running 40 times faster.
17.Noise2Void - Learning Denoising from Single Noisy Images pdf
The field of image denoising is currently dominated by discriminative deep learning methods that are trained on pairs of noisy input and clean target images. Recently it has been shown that such methods can also be trained without clean targets. Instead, independent pairs of noisy images can be used, in an approach known as Noise2Noise (N2N). Here, we introduce Noise2Void (N2V), a training scheme that takes this idea one step further. It does not require noisy image pairs, nor clean target images. Consequently, N2V allows us to train directly on the body of data to be denoised and can therefore be applied when other methods cannot. Especially interesting is the application to biomedical image data, where the acquisition of training targets, clean or noisy, is frequently not possible. We compare the performance of N2V to approaches that have either clean target images and/or noisy image pairs available. Intuitively, N2V cannot be expected to outperform methods that have more information available during training. Still, we observe that the denoising performance of Noise2Void drops in moderation and compares favorably to training-free denoising methods.
18.One-Shot Item Search with Multimodal Data pdf
In the task of near similar image search, features from Deep Neural Network is often used to compare images and measure similarity. In the past, we only focused visual search in image dataset without text data. However, since deep neural network emerged, the performance of visual search becomes high enough to apply it in many industries from 3D data to multimodal data. Compared to the needs of multimodal search, there has not been sufficient researches. In this paper, we present a method of near similar search with image and text multimodal dataset. Earlier time, similar image search, especially when searching shopping items, treated image and text separately to search similar items and reorder the results. This regards two tasks of image search and text matching as two different tasks. Our method, however, explore the vast data to compute k-nearest neighbors using both image and text. In our experiment of similar item search, our system using multimodal data shows better performance than single data while it only increases minute computing time. For the experiment, we collected more than 15 million of accessory and six million of digital product items from online shopping websites, in which the product item comprises item images, titles, categories, and descriptions. Then we compare the performance of multimodal searching to single space searching in these datasets.
19.MagicVO: End-to-End Monocular Visual Odometry through Deep Bi-directional Recurrent Convolutional Neural Network pdf
This paper proposes a new framework to solve the problem of monocular visual odometry, called MagicVO . Based on Convolutional Neural Network (CNN) and Bi-directional LSTM (Bi-LSTM), MagicVO outputs a 6-DoF absolute-scale pose at each position of the camera with a sequence of continuous monocular images as input. It not only utilizes the outstanding performance of CNN in image feature processing to extract the rich features of image frames fully but also learns the geometric relationship from image sequences pre and post through Bi-LSTM to get a more accurate prediction. A pipeline of the MagicVO is shown in Fig. 1. The MagicVO system is end-to-end, and the results of experiments on the KITTI dataset and the ETH-asl cla dataset show that MagicVO has a better performance than traditional visual odometry (VO) systems in the accuracy of pose and the generalization ability.
20.Deep Learned Frame Prediction for Video Compression pdf
Motion compensation is one of the most essential methods for any video compression algorithm. Video frame prediction is a task analogous to motion compensation. In recent years, the task of frame prediction is undertaken by deep neural networks (DNNs). In this thesis we create a DNN to perform learned frame prediction and additionally implement a codec that contains our DNN. We train our network using two methods for two different goals. Firstly we train our network based on mean square error (MSE) only, aiming to obtain highest PSNR values at frame prediction and video compression. Secondly we use adversarial training to produce visually more realistic frame predictions. For frame prediction, we compare our method with the baseline methods of frame difference and 16x16 block motion compensation. For video compression we further include x264 video codec in the comparison. We show that in frame prediction, adversarial training produces frames that look sharper and more realistic, compared MSE based training, but in video compression it consistently performs worse. This proves that even though adversarial training is useful for generating video frames that are more pleasing to the human eye, they should not be employed for video compression. Moreover, our network trained with MSE produces accurate frame predictions, and in quantitative results, for both tasks, it produces comparable results in all videos and outperforms other methods on average. More specifically, learned frame prediction outperforms other methods in terms of rate-distortion performance in case of high motion video, while the rate-distortion performance of our method is competitive with that of x264 in low motion video.
21.Deep Geometric Prior for Surface Reconstruction pdf
The reconstruction of a discrete surface from a point cloud is a fundamental geometry processing problem that has been studied for decades, with many methods developed. We propose the use of a deep neural network as a geometric prior for surface reconstruction. Specifically, we overfit a neural network representing a local chart parameterization to part of an input point cloud using the Wasserstein distance as a measure of approximation. By jointly fitting many such networks to overlapping parts of the point cloud, while enforcing a consistency condition, we compute a manifold atlas. By sampling this atlas, we can produce a dense reconstruction of the surface approximating the input cloud. The entire procedure does not require any training data or explicit regularization, yet, we show that it is able to perform remarkably well: not introducing typical overfitting artifacts, and approximating sharp features closely at the same time. We experimentally show that this geometric prior produces good results for both man-made objects containing sharp features and smoother organic objects, as well as noisy inputs. We compare our method with a number of well-known reconstruction methods on a standard surface reconstruction benchmark.
22.Beyond One Glance: Gated Recurrent Architecture for Hand Segmentation pdf
As mixed reality is gaining increased momentum, the development of effective and efficient solutions to egocentric hand segmentation is becoming critical. Traditional segmentation techniques typically follow a one-shot approach, where the image is passed forward only once through a model that produces a segmentation mask. This strategy, however, does not reflect the perception of humans, who continuously refine their representation of the world. In this paper, we therefore introduce a novel gated recurrent architecture. It goes beyond both iteratively passing the predicted segmentation mask through the network and adding a standard recurrent unit to it. Instead, it incorporates multiple encoder-decoder layers of the segmentation network, so as to keep track of its internal state in the refinement process. As evidenced by our results on standard hand segmentation benchmarks and on our own dataset, our approach outperforms these other, simpler recurrent segmentation techniques, as well as the state-of-the-art hand segmentation one. Furthermore, we demonstrate the generality of our approach by applying it to road segmentation, where it also outperforms other baseline methods.
23.Efficient Image Retrieval via Decoupling Diffusion into Online and Offline Processing pdf
Diffusion is commonly used as a ranking or re-ranking method in retrieval tasks to achieve higher retrieval performance, and has attracted lots of attention in recent years. A downside to diffusion is that it performs slowly in comparison to the naive k-NN search, which causes a non-trivial online computational cost on large datasets. To overcome this weakness, we propose a novel diffusion technique in this paper. In our work, instead of applying diffusion to the query, we pre-compute the diffusion results of each element in the database, making the online search a simple linear combination on top of the k-NN search process. Our proposed method becomes 10~ times faster in terms of online search speed. Moreover, we propose to use late truncation instead of early truncation in previous works to achieve better retrieval performance.
24.Are 2D-LSTM really dead for offline text recognition? pdf
There is a recent trend in handwritten text recognition with deep neural networks to replace 2D recurrent layers with 1D, and in some cases even completely remove the recurrent layers, relying on simple feed-forward convolutional only architectures. The most used type of recurrent layer is the Long-Short Term Memory (LSTM). The motivations to do so are many: there are few open-source implementations of 2D-LSTM, even fewer supporting GPU implementations (currently cuDNN only implements 1D-LSTM); 2D recurrences reduce the amount of computations that can be parallelized, and thus possibly increase the training/inference time; recurrences create global dependencies with respect to the input, and sometimes this may not be desirable.
Many recent competitions were won by systems that employed networks that use 2D-LSTM layers. Most previous work that compared 1D or pure feed-forward architectures to 2D recurrent models have done so on simple datasets or did not fully optimize the "baseline" 2D model compared to the challenger model, which was dully optimized.
In this work, we aim at a fair comparison between 2D and competing models and also extensively evaluate them on more complex datasets that are more representative of challenging "real-world" data, compared to "academic" datasets that are more restricted in their complexity. We aim at determining when and why the 1D and 2D recurrent models have different results. We also compare the results with a language model to assess if linguistic constraints do level the performance of the different networks.
Our results show that for challenging datasets, 2D-LSTM networks still seem to provide the highest performances and we propose a visualization strategy to explain it.
25.DSBI: Double-Sided Braille Image Dataset and Algorithm Evaluation for Braille Dots Detection pdf
Braille is an effective way for the visually impaired to learn knowledge and obtain information. Braille image recognition aims to automatically detect Braille dots in the whole Braille image. There is no available public datasets for Braille image recognition to push relevant research and evaluate algorithms. This paper constructs a large-scale Double-Sided Braille Image dataset DSBI with detailed Braille recto dots, verso dots and Braille cells annotation. To quickly annotate Braille images, an auxiliary annotation strategy is proposed, which adopts initial automatic detection of Braille dots and modifies annotation results by convenient human-computer interaction method. This labeling strategy can averagely increase label efficiency by six times for recto dots annotation in one Braille image. Braille dots detection is the core and basic step for Braille image recognition. This paper also evaluates some Braille dots detection methods on our dataset DSBI and gives the benchmark performance of recto dots detection. We have released our Braille images dataset on the GitHub website.
26.UnDEMoN 2.0: Improved Depth and Ego Motion Estimation through Deep Image Sampling pdf
In this paper, we provide an improved version of UnDEMoN model for depth and ego motion estimation from monocular images. The improvement is achieved by combining the standard bi-linear sampler with a deep network based image sampling model (DIS-NET) to provide better image reconstruction capabilities on which the depth estimation accuracy depends in un-supervised learning models. While DIS-NET provides higher order regression and larger input search space, the bi-linear sampler provides geometric constraints necessary for reducing the size of the solution space for an ill-posed problem of this kind. This combination is shown to provide significant improvement in depth and pose estimation accuracy outperforming all existing state-of-the-art methods in this category. In addition, the modified network uses far less number of tunable parameters making it one of the lightest deep network model for depth estimation. The proposed model is labeled as "UnDEMoN 2.0" indicating an improvement over the existing UnDEMoN model. The efficacy of the proposed model is demonstrated through rigorous experimental analysis on the standard KITTI dataset.
27.Automatic Image Stylization Using Deep Fully Convolutional Networks pdf
Color and tone stylization strives to enhance unique themes with artistic color and tone adjustments. It has a broad range of applications from professional image postprocessing to photo sharing over social networks. Mainstream photo enhancement softwares provide users with predefined styles, which are often hand-crafted through a trial-and-error process. Such photo adjustment tools lack a semantic understanding of image contents and the resulting global color transform limits the range of artistic styles it can represent. On the other hand, stylistic enhancement needs to apply distinct adjustments to various semantic regions. Such an ability enables a broader range of visual styles. In this paper, we propose a novel deep learning architecture for automatic image stylization, which learns local enhancement styles from image pairs. Our deep learning architecture is an end-to-end deep fully convolutional network performing semantics-aware feature extraction as well as automatic image adjustment prediction. Image stylization can be efficiently accomplished with a single forward pass through our deep network. Experiments on existing datasets for image stylization demonstrate the effectiveness of our deep learning architecture.
28.Affinity Derivation and Graph Merge for Instance Segmentation pdf
We present an instance segmentation scheme based on pixel affinity information, which is the relationship of two pixels belonging to a same instance. In our scheme, we use two neural networks with similar structure. One is to predict pixel level semantic score and the other is designed to derive pixel affinities.
Regarding pixels as the vertexes and affinities as edges, we then propose a simple yet effective graph merge algorithm to cluster pixels into instances. Experimental results show that our scheme can generate fine-grained instance mask.
With Cityscapes training data, the proposed scheme achieves 27.3 AP on test set.
29.Object Tracking by Reconstruction with View-Specific Discriminative Correlation Filters pdf
Standard RGB-D trackers treat the target as an inherently 2D structure, which makes modelling appearance changes related even to simple out-of-plane rotation highly challenging. We address this limitation by proposing a novel long-term RGB-D tracker - Object Tracking by Reconstruction (OTR). The tracker performs online 3D target reconstruction to facilitate robust learning of a set of view-specific discriminative correlation filters (DCFs). The 3D reconstruction supports two performance-enhancing features: (i) generation of accurate spatial support for constrained DCF learning from its 2D projection and (ii) point cloud based estimation of 3D pose change for selection and storage of view-specific DCFs which are used to robustly localize the target after out-of-view rotation or heavy occlusion. Extensive evaluation of OTR on the challenging Princeton RGB-D tracking and STC Benchmarks shows it outperforms the state-of-the-art by a large margin.
30.Sampling Techniques for Large-Scale Object Detection from Sparsely Annotated Objects pdf
Efficient and reliable methods for training of object detectors are in higher demand than ever, and more and more data relevant to the field is becoming available. However, large datasets like Open Images Dataset v4 (OID) are sparsely annotated, and some measure must be taken in order to ensure the training of a reliable detector. In order to take the incompleteness of these datasets into account, one possibility is to use pretrained models to detect the presence of the unverified objects. However, the performance of such a strategy depends largely on the power of the pretrained model. In this study, we propose part-aware sampling, a method that uses human intuition for the hierarchical relation between objects. In terse terms, our method works by making assumptions like "a bounding box for a car should contain a bounding box for a tire". We demonstrate the power of our method on OID and compare the performance against a method based on a pretrained model. Our method also won the first and second place on the public and private test sets of the Google AI Open Images Competition 2018.
31.Algae Detection Using Computer Vision and Deep Learning pdf
A disconcerting ramification of water pollution caused by burgeoning populations, rapid industrialization and modernization of agriculture, has been the exponential increase in the incidence of algal growth across the globe. Harmful algal blooms (HABs) have devastated fisheries, contaminated drinking water and killed livestock, resulting in economic losses to the tune of millions of dollars. Therefore, it is important to constantly monitor water bodies and identify any algae build-up so that prompt action against its accumulation can be taken and the harmful consequences can be avoided. In this paper, we propose a computer vision system based on deep learning for algae monitoring. The proposed system is fast, accurate and cheap, and it can be installed on any robotic platforms such as USVs and UAVs for autonomous algae monitoring. The experimental results demonstrate that the proposed system can detect algae in distinct environments regardless of the underlying hardware with high accuracy and in real time.
32.CIAN: Cross-Image Affinity Net for Weakly Supervised Semantic Segmentation pdf
Weakly supervised semantic segmentation based on image-level labels aims for alleviating the data scarcity problem by training with coarse labels. State-of-the-art methods rely on image-level labels to generate proxy segmentation masks, then train the segmentation network on these masks with various constraints. These methods consider each image independently and lack the exploration of cross-image relationships. We argue the cross-image relationship is vital to weakly supervised learning. We propose an end-to-end affinity module for explicitly modeling the relationship among a group of images. By means of this, one image can benefit from the complementary information from other images, and the supervision guidance can be shared in the group. The proposed method improves over the baseline with a large margin. Our method achieves 64.1% mIOU score on Pascal VOC 2012 validation set, and 64.7% mIOU score on test set, which is a new state-of-the-art by only using image-level labels, demonstrating the effectiveness of the method.
33.Part-level Car Parsing and Reconstruction from Single Street View pdf
In this paper, we make the first attempt to build a framework to simultaneously estimate semantic parts, shape, translation, and orientation of cars from single street view. Our framework contains three major contributions. Firstly, a novel domain adaptation approach based on the class consistency loss is developed to transfer our part segmentation model from the synthesized images to the real images. Secondly, we propose a novel network structure that leverages part-level features from street views and 3D losses for pose and shape estimation. Thirdly, we construct a high quality dataset that contains more than 300 different car models with physical dimensions and part-level annotations based on global and local deformations. We have conducted experiments on both synthesized data and real images. Our results show that the domain adaptation approach can bring 35.5 percentage point performance improvement in terms of mean intersection-over-union score (mIoU) comparing with the baseline network using domain randomization only. Our network for translation and orientation estimation achieves competitive performance on highly complex street views (e.g., 11 cars per image on average). Moreover, our network is able to reconstruct a list of 3D car models with part-level details from street views, which could benefit various applications such as fine-grained car recognition, vehicle re-identification, and traffic simulation.
34.From Recognition to Cognition: Visual Commonsense Reasoning pdf
Visual understanding goes well beyond object recognition. With one glance at an image, we can effortlessly imagine the world beyond the pixels: for instance, we can infer people's actions, goals, and mental states. While this task is easy for humans, it is tremendously difficult for today's vision systems, requiring higher-order cognition and commonsense reasoning about the world. In this paper, we formalize this task as Visual Commonsense Reasoning. In addition to answering challenging visual questions expressed in natural language, a model must provide a rationale explaining why its answer is true. We introduce a new dataset, VCR, consisting of 290k multiple choice QA problems derived from 110k movie scenes. The key recipe to generating non-trivial and high-quality problems at scale is Adversarial Matching, a new approach to transform rich annotations into multiple choice questions with minimal bias. To move towards cognition-level image understanding, we present a new reasoning engine, called Recognition to Cognition Networks (R2C), that models the necessary layered inferences for grounding, contextualization, and reasoning. Experimental results show that while humans find VCR easy (over 90% accuracy), state-of-the-art models struggle (~45%). Our R2C helps narrow this gap (~65%); still, the challenge is far from solved, and we provide analysis that suggests avenues for future work.
35.Noise-tolerant Audio-visual Online Person Verification using an Attention-based Neural Network Fusion pdf
In this paper, we present a multi-modal online person verification system using both speech and visual signals. Inspired by neuroscientific findings on the association of voice and face, we propose an attention-based end-to-end neural network that learns multi-sensory associations for the task of person verification. The attention mechanism in our proposed network learns to conditionally select a salient modality between speech and facial representations that provides a balance between complementary inputs. By virtue of this capability, the network is robust to missing or corrupted data from either modality. In the VoxCeleb2 dataset, we show that our method performs favorably against competing multi-modal methods. Even for extreme cases of large corruption or an entirely missing modality, our method demonstrates robustness over other unimodal methods.
36.Perceptual Conditional Generative Adversarial Networks for End-to-End Image Colourization pdf
Colours are everywhere. They embody a significant part of human visual perception. In this paper, we explore the paradigm of hallucinating colours from a given gray-scale image. The problem of colourization has been dealt in previous literature but mostly in a supervised manner involving user-interference. With the emergence of Deep Learning methods numerous tasks related to computer vision and pattern recognition have been automatized and carried in an end-to-end fashion due to the availability of large data-sets and high-power computing systems. We investigate and build upon the recent success of Conditional Generative Adversarial Networks (cGANs) for Image-to-Image translations. In addition to using the training scheme in the basic cGAN, we propose an encoder-decoder generator network which utilizes the class-specific cross-entropy loss as well as the perceptual loss in addition to the original objective function of cGAN. We train our model on a large-scale dataset and present illustrative qualitative and quantitative analysis of our results. Our results vividly display the versatility and proficiency of our methods through life-like colourization outcomes.
37.Probability-based Detection Quality (PDQ): A Probabilistic Approach to Detection Evaluation pdf
We propose a new visual object detector evaluation measure which not only assesses detection quality, but also accounts for the spatial and label uncertainties produced by object detection systems. Current evaluation measures such as mean average precision (mAP) do not take these two aspects into account, accepting detections with no spatial uncertainty and using only the label with the winning score instead of a full class probability distribution to rank detections. To overcome these limitations, we propose the probability-based detection quality (PDQ) measure which evaluates both spatial and label probabilities, requires no thresholds to be predefined, and optimally assigns ground-truth objects to detections. Our experimental evaluation shows that PDQ rewards detections with accurate spatial probabilities and explicitly evaluates label probability to determine detection quality. PDQ aims to encourage the development of new object detection approaches that provide meaningful spatial and label uncertainty measures.
38.A Fully Sequential Methodology for Convolutional Neural Networks pdf
Recent work has shown that the performance of convolutional neural networks could be significantly improved by increasing the depth of the representation. We propose a fully sequential methodology to construct and train extremely deep convolutional neural networks.
We first introduce a novel sequential convolutional layer to construct the network. The proposed layer is capable of constructing trainable and highly efficient feedforward networks that consist of thousands of vanilla convolutional layers with rather limited number of parameters. The layer extracts each feature of the produced representation in sequence, allowing feature reuse within the layer. This form of feature reuse introduces in-layer hierarchy to the extracted features which greatly increases the depth of the representation, enabling richer structures to be explored.
Furthermore, we employ the progressive growing training method to optimize each module of the network in sequence. This training manner progressively increases the network capacity allowing later modules to be optimized conditioning on prior knowledge from earlier modules. Thus, it encourages long term dependency to be established among each module of the network, which increases the effective depth of networks with skip connections, as well alleviates multiple optimization difficulties for deep networks.
39.Reconstruction Loss Minimized FCN for Single Image Dehazing pdf
Haze and fog reduce the visibility of outdoor scenes as a veil like semi-transparent layer appears over the objects. As a result, images captured under such conditions lack contrast. Image dehazing methods try to alleviate this problem by recovering a clear version of the image. In this paper, we propose a Fully Convolutional Neural Network based model to recover the clear scene radiance by estimating the environmental illumination and the scene transmittance jointly from a hazy image. The method uses a relaxed haze imaging model to allow for the situations with non-uniform illumination. We have trained the network by minimizing a custom-defined loss that measures the error of reconstructing the hazy image in three different ways. Additionally, we use a multilevel approach to determine the scene transmittance and the environmental illumination in order to reduce the dependence of the estimate on image scale. Evaluations show that our model performs well compared to the existing state-of-the-art methods. It also verifies the potential of our model in diverse situations and various lighting conditions.
40.Unsupervised Image Captioning pdf
Deep neural networks have achieved great successes on the image captioning task. However, most of the existing models depend heavily on paired image-sentence datasets, which are very expensive to acquire. In this paper, we make the first attempt to train an image captioning model in an unsupervised manner. Instead of relying on manually labeled image-sentence pairs, our proposed model merely requires an image set, a sentence corpus, and an existing visual concept detector. The sentence corpus is used to teach the captioning model how to generate plausible sentences. Meanwhile, the knowledge in the visual concept detector is distilled into the captioning model to guide the model to recognize the visual concepts in an image. In order to further encourage the generated captions to be semantically consistent with the image, the image and caption are projected into a common latent space so that they can be used to reconstruct each other. Given that the existing sentence corpora are mainly designed for linguistic research and thus with little reference to image contents, we crawl a large-scale image description corpus of 2 million natural sentences to facilitate the unsupervised image captioning scenario. Experimental results show that our proposed model is able to produce quite promising results without using any labeled training pairs.
41.Tackling Early Sparse Gradients in Softmax Activation Using Leaky Squared Euclidean Distance pdf
Softmax activation is commonly used to output the probability distribution over categories based on certain distance metric. In scenarios like one-shot learning, the distance metric is often chosen to be squared Euclidean distance between the query sample and the category prototype. This practice works well in most time. However, we find that choosing squared Euclidean distance may cause distance explosion leading gradients to be extremely sparse in the early stage of back propagation. We term this phenomena as the early sparse gradients problem. Though it doesn't deteriorate the convergence of the model, it may set up a barrier to further model improvement. To tackle this problem, we propose to use leaky squared Euclidean distance to impose a restriction on distances. In this way, we can avoid distance explosion and increase the magnitude of gradients. Extensive experiments are conducted on Omniglot and miniImageNet datasets. We show that using leaky squared Euclidean distance can improve one-shot classification accuracy on both datasets.
42.Event-Based Structured Light for Depth Reconstruction using Frequency Tagged Light Patterns pdf
This paper presents a new method for 3D depth estimation using the output of an asynchronous time driven image sensor. In association with a high speed Digital Light Processing projection system, our method achieves real-time reconstruction of 3D points cloud, up to several hundreds of hertz. Unlike state of the art methodology, we introduce a method that relies on the use of frequency tagged light pattern that make use of the high temporal resolution of event based sensors. This approch eases matching as each pattern unique frequency allow for any easy matching between displayed patterns and the event based sensor. Results are show on real scenes.
43.Generating Attention from Classifier Activations for Fine-grained Recognition pdf
Recent advances in fine-grained recognition utilize attention maps to localize objects of interest. Although there are many ways to generate attention maps, most of them rely on sophisticated loss functions or complex training processes. In this work, we propose a simple and straightforward attention generation model based on the output activations of classifiers. The advantage of our model is that it can be easily trained with image level labels and softmax loss functions. More specifically, multiple linear local classifiers are firstly adopted to perform fine-grained classification at each location of high level CNN feature maps. The attention map is generated by aggregating and max-pooling the output activations. Then the attention map serves as a surrogate target object mask to train those local classifiers, similar to training models for semantic segmentation. Our model achieves state-of-the-art results on three heavily benchmarked datasets, i.e. 87.9% on CUB-200-2011 dataset, 94.1% on Stanford Cars dataset and 92.1% on FGVC-Aircraft dataset, demonstrating its effectiveness on fine-grained recognition tasks.
44.Quality-Aware Multimodal Saliency Detection via Deep Reinforcement Learning pdf
Incorporating various modes of information into the machine learning procedure is becoming a new trend. And data from various source can provide more information than single one no matter they are heterogeneous or homogeneous. Existing deep learning based algorithms usually directly concatenate features from each domain to represent the input data. Seldom of them take the quality of data into consideration which is a key issue in related multimodal problems. In this paper, we propose an efficient quality-aware deep neural network to model the weight of data from each domain using deep reinforcement learning (DRL). Specifically, we take the weighting of each domain as a decision-making problem and teach an agent learn to interact with the environment. The agent can tune the weight of each domain through discrete action selection and obtain a positive reward if the saliency results are improved. The target of the agent is to achieve maximum rewards after finished its sequential action selection. We validate the proposed algorithms on multimodal saliency detection in a coarse-to-fine way. The coarse saliency maps are generated from an encoder-decoder framework which is trained with content loss and adversarial loss. The final results can be obtained via adaptive weighting of maps from each domain. Experiments conducted on two kinds of salient object detection benchmarks validated the effectiveness of our proposed quality-aware deep neural network.
45.A Coarse-to-fine Deep Convolutional Neural Network Framework for Frame Duplication Detection and Localization in Video Forgery pdf
Frame duplication is to duplicate a sequence of consecutive frames and insert or replace to conceal or imitate a specific event/content in the same source video. To automatically detect the duplicated frames in a manipulated video, we propose a coarse-to-fine deep convolutional neural network framework to detect and localize the frame duplications. We first run an I3D network to obtain the most candidate duplicated frame sequences and selected frame sequences, and then run a Siamese network with ResNet network to identify each pair of a duplicated frame and the corresponding selected frame. We also propose a heuristic strategy to formulate the video-level score. We then apply our inconsistency detector fine-tuned on the I3D network to distinguish duplicated frames from selected frames. With the experimental evaluation conducted on two video datasets, we strongly demonstrate that our proposed method outperforms the current state-of-the-art methods.
46.Joint Monocular 3D Vehicle Detection and Tracking pdf
3D vehicle detection and tracking from a monocular camera requires detecting and associating vehicles, and estimating their locations and extents together. It is challenging because vehicles are in constant motion and it is practically impossible to recover the 3D positions from a single image. In this paper, we propose a novel framework that jointly detects and tracks 3D vehicle bounding boxes. Our approach leverages 3D pose estimation to learn 2D patch association overtime and uses temporal information from tracking to obtain stable 3D estimation. Our method also leverages 3D box depth ordering and motion to link together the tracks of occluded objects. We train our system on realistic 3D virtual environments, collecting a new diverse, large-scale and densely annotated dataset with accurate 3D trajectory annotations. Our experiments demonstrate that our method benefits from inferring 3D for both data association and tracking robustness, leveraging our dynamic 3D tracking dataset.
47.MIST: Multiple Instance Spatial Transformer Network pdf
We propose a deep network that can be trained to tackle image reconstruction and classification problems that involve detection of multiple object instances, without any supervision regarding their whereabouts. The network learns to extract the most significant top-K patches, and feeds these patches to a task-specific network -- e.g., auto-encoder or classifier -- to solve a domain specific problem. The challenge in training such a network is the non-differentiable top-K selection process. To address this issue, we lift the training optimization problem by treating the result of top-K selection as a slack variable, resulting in a simple, yet effective, multi-stage training. Our method is able to learn to detect recurrent structures in the training dataset by learning to reconstruct images. It can also learn to localize structures when only knowledge on the occurrence of the object is provided, and in doing so it outperforms the state-of-the-art.
48.IGNOR: Image-guided Neural Object Rendering pdf
We propose a new learning-based novel view synthesis approach for scanned objects that is trained based on a set of multi-view images. Instead of using texture mapping or hand-designed image-based rendering, we directly train a deep neural network to synthesize a view-dependent image of an object. First, we employ a coverage-based nearest neighbour look-up to retrieve a set of reference frames that are explicitly warped to a given target view using cross-projection. Our network then learns to best composite the warped images. This enables us to generate photo-realistic results, while not having to allocate capacity on `remembering' object appearance. Instead, the multi-view images can be reused. While this works well for diffuse objects, cross-projection does not generalize to view-dependent effects. Therefore, we propose a decomposition network that extracts view-dependent effects and that is trained in a self-supervised manner. After decomposition, the diffuse shading is cross-projected, while the view-dependent layer of the target view is regressed. We show the effectiveness of our approach both qualitatively and quantitatively on real as well as synthetic data.
49.Learning View Priors for Single-view 3D Reconstruction pdf
There is some ambiguity in the 3D shape of an object when the number of observed views is small. Because of this ambiguity, although a 3D object reconstructor can be trained using a single view or a few views per object, reconstructed shapes only fit the observed views and appear incorrect from the unobserved viewpoints. To reconstruct shapes that look reasonable from any viewpoint, we propose to train a discriminator that learns prior knowledge regarding possible views. The discriminator is trained to distinguish the reconstructed views of the observed viewpoints from those of the unobserved viewpoints. The reconstructor is trained to correct unobserved views by fooling the discriminator. Our method outperforms current state-of-the-art methods on both synthetic and natural image datasets; this validates the effectiveness of our method.
50.Bilateral Adversarial Training: Towards Fast Training of More Robust Models Against Adversarial Attacks pdf
In this paper, we study fast training of adversarially robust models. From the analyses on the state-of-the-art defense method, i.e., the multi-step adversarial training~\cite{madry2017towards}, we hypothesize that the gradient magnitude links to the model robustness. Motivated by this, we propose to perturb both the image and the label during training, which we call Bilateral Adversarial Training (BAT). To generate the adversarial label, we derive an closed-form heuristic solution. To generate the adversarial image, we use one-step targeted attack with the target label being the most confusing class. In the experiment, we first show that random start and the most confusing target attack effectively prevent the label leaking and gradient masking problem. Then coupled with the adversarial label part, our model significantly improves the state-of-the-art results. For example, against PGD100 attack with cross-entropy loss, on CIFAR10, we achieve 63.7% versus 47.2%; on SVHN, we achieve 59.1% versus 42.1%; on CIFAR100, we achieve 25.3% versus 23.4%. Note that these results are obtained by the fast one-step adversarial training.
51.Time-Aware and View-Aware Video Rendering for Unsupervised Representation Learning pdf
The recent success in deep learning has lead to various effective representation learning methods for videos. However, the current approaches for video representation require large amount of human labeled datasets for effective learning. We present an unsupervised representation learning framework to encode scene dynamics in videos captured from multiple viewpoints. The proposed framework has two main components: Representation Learning Network (RL-NET), which learns a representation with the help of Blending Network (BL-NET), and Video Rendering Network (VR-NET), which is used for video synthesis. The framework takes as input video clips from different viewpoints and time, learns an internal representation and uses this representation to render a video clip from an arbitrary given viewpoint and time. The ability of the proposed network to render video frames from arbitrary viewpoints and time enable it to learn a meaningful and robust representation of the scene dynamics. We demonstrate the effectiveness of the proposed method in rendering view-aware as well as time-aware video clips on two different real-world datasets including UCF-101 and NTU-RGB+D. To further validate the effectiveness of the learned representation, we use it for the task of view-invariant activity classification where we observe a significant improvement (~26%) in the performance on NTU-RGB+D dataset compared to the existing state-of-the art methods.
52.LSTA: Long Short-Term Attention for Egocentric Action Recognition pdf
Egocentric activity recognition is one of the most challenging tasks in video analysis. It requires a fine-grained discrimination of small objects and their manipulation. While some methods base on strong supervision and attention mechanisms, they are either annotation consuming or do not take spatio-temporal patterns into account. In this paper we propose LSTA as a mechanism to focus on features from spatial relevant parts while attention is being tracked smoothly across the video sequence. We demonstrate the effectiveness of LSTA on egocentric activity recognition with an end-to-end trainable two-stream architecture, achieving state of the art performance on four standard benchmarks.
53.Attentive Relational Networks for Mapping Images to Scene Graphs pdf
Scene graph generation refers to the task of automatically mapping an image into a semantic structural graph, which requires correctly labeling each extracted objects and their interaction relationships. Despite the recent successes in object detection using deep learning techniques, inferring complex contextual relationships and structured graph representations from visual data remains a challenging topic. In this study, we propose a novel Attentive Relational Network that consists of two key modules with an object detection backbone to approach this problem. The first module is a semantic transformation module used to capture semantic embedded relation features, by translating visual features and linguistic features into a common semantic space. The other module is a graph self-attention module introduced to embed a joint graph representation through assigning various importance weights to neighboring nodes. Finally, accurate scene graphs are produced with the relation inference module by recognizing all entities and the corresponding relations. We evaluate our proposed method on the widely-adopted Visual Genome Dataset, and the results demonstrate the effectiveness and superiority of our model.
54.Matching Features without Descriptors: Implicitly Matched Interest Points (IMIPs) pdf
The extraction and matching of interest points is a prerequisite for visual pose estimation and related problems. Traditionally, matching has been achieved by assigning descriptors to interest points and matching points that have similar descriptors. In this paper, we propose a method by which interest points are instead already implicitly matched at detection time. Thanks to this, descriptors do not need to be calculated, stored, communicated, or matched any more. This is achieved by a convolutional neural network with multiple output channels. The i-th interest point is the location of the maximum of the i-th channel, and the i-th interest point in one image is implicitly matched with the i-th interest point in another image. This paper describes how to design and train such a network in a way that results in successful relative pose estimation performance with as little as 128 output channels. While the overall matching score is slightly lower than with traditional methods, the network also outputs the confidence for a specific interest point resulting in a valid match. Most importantly, the approach completely gets rid of descriptors and thus enables localization systems with a significantly smaller memory footprint and multi-agent localization systems that require significantly less bandwidth. We evaluate performance relative to state-of-the-art alternatives.
55.GANsfer Learning: Combining labelled and unlabelled data for GAN based data augmentation pdf
Medical imaging is a domain which suffers from a paucity of manually annotated data for the training of learning algorithms. Manually delineating pathological regions at a pixel level is a time consuming process, especially in 3D images, and often requires the time of a trained expert. As a result, supervised machine learning solutions must make do with small amounts of labelled data, despite there often being additional unlabelled data available. Whilst of less value than labelled images, these unlabelled images can contain potentially useful information. In this paper we propose combining both labelled and unlabelled data within a GAN framework, before using the resulting network to produce images for use when training a segmentation network. We explore the task of deep grey matter multi-class segmentation in an AD dataset and show that the proposed method leads to a significant improvement in segmentation results, particularly in cases where the amount of labelled data is restricted. We show that this improvement is largely driven by a greater ability to segment the structures known to be the most affected by AD, thereby demonstrating the benefits of exposing the system to more examples of pathological anatomical variation. We also show how a shift in domain of the training data from young and healthy towards older and more pathological examples leads to better segmentations of the latter cases, and that this leads to a significant improvement in the ability for the computed segmentations to stratify cases of AD.
56.Art2Real: Unfolding the Reality of Artworks via Semantically-Aware Image-to-Image Translation pdf
The applicability of computer vision to real paintings and artworks has been rarely investigated, even though a vast heritage would greatly benefit from techniques which can understand and process data from the artistic domain. This is partially due to the small amount of annotated artistic data, which is not even comparable to that of natural images captured by cameras. In this paper, we propose a semantic-aware architecture which can translate artworks to photo-realistic visualizations, thus reducing the gap between visual features of artistic and realistic data. Our architecture can generate natural images by retrieving and learning details from real photos through a similarity matching strategy which leverages a weakly-supervised semantic understanding of the scene. Experimental results show that the proposed technique leads to increased realism and to a reduction in domain shift, which improves the performance of pre-trained architectures for classification, detection, and segmentation. Code will be made publicly available.
57.Show, Control and Tell: A Framework for Generating Controllable and Grounded Captions pdf
Current captioning approaches can describe images using black-box architectures whose behavior is hardly controllable and explainable from the exterior. As an image can be described in infinite ways depending on the goal and the context at hand, a higher degree of controllability is needed to apply captioning algorithms in complex scenarios. In this paper, we introduce a novel framework for image captioning which can generate diverse descriptions by allowing both grounding and controllability. Given a control signal in the form of a sequence or set of image regions, we generate the corresponding caption through a recurrent architecture which predicts textual chunks explicitly grounded on regions, following the constraints of the given control. Experiments are conducted on Flickr30k Entities and on COCO Entities, an extended version of COCO in which we add grounding annotations collected in a semi-automatic manner. Results demonstrate that our method achieves state of the art performances on controllable image captioning, in terms of caption quality and diversity. Code will be made publicly available.
58.Understanding Image Quality and Trust in Peer-to-Peer Marketplaces pdf
As any savvy online shopper knows, second-hand peer-to-peer marketplaces are filled with images of mixed quality. How does image quality impact marketplace outcomes, and can quality be automatically predicted? In this work, we conducted a large-scale study on the quality of user-generated images in peer-to-peer marketplaces. By gathering a dataset of common second-hand products (~75,000 images) and annotating a subset with human-labeled quality judgments, we were able to model and predict image quality with decent accuracy (~87%). We then conducted two studies focused on understanding the relationship between these image quality scores and two marketplace outcomes: sales and perceived trustworthiness. We show that image quality is associated with higher likelihood that an item will be sold, though other factors such as view count were better predictors of sales. Nonetheless, we show that high quality user-generated images selected by our models outperform stock imagery in eliciting perceptions of trust from users. Our findings can inform the design of future marketplaces and guide potential sellers to take better product images.
59.Evolving Space-Time Neural Architectures for Videos pdf
In this paper, we present a new method for evolving video CNN models to find architectures that more optimally captures rich spatio-temporal information in videos. Previous work, taking advantage of 3D convolutional layers, obtained promising results by manually designing CNN architectures for videos. We here develop an evolutionary algorithm that automatically explores models with different types and combinations of space-time convolutional layers to jointly capture various spatial and temporal aspects of video representations. We further propose a new key component in video model evolution, the iTGM layer, which more efficiently utilizes its parameters to allow learning of space-time interactions over longer time horizons. The experiments confirm the advantages of our video CNN architecture evolution, with results outperforming previous state-of-the-art models. Our algorithm discovers new and interesting video architecture structures.
60.Learning Robust Representations for Automatic Target Recognition pdf
Radio frequency (RF) sensors are used alongside other sensing modalities to provide rich representations of the world. Given the high variability of complex-valued target responses, RF systems are susceptible to attacks masking true target characteristics from accurate identification. In this work, we evaluate different techniques for building robust classification architectures exploiting learned physical structure in received synthetic aperture radar signals of simulated 3D targets.
61.Adversarial Video Compression Guided by Soft Edge Detection pdf
We propose a video compression framework using conditional Generative Adversarial Networks (GANs). We rely on two encoders: one that deploys a standard video codec and another which generates low-level maps via a pipeline of down-sampling, a newly devised soft edge detector, and a novel lossless compression scheme. For decoding, we use a standard video decoder as well as a neural network based one, which is trained using a conditional GAN. Recent "deep" approaches to video compression require multiple videos to pre-train generative networks to conduct interpolation. In contrast to this prior work, our scheme trains a generative decoder on pairs of a very limited number of key frames taken from a single video and corresponding low-level maps. The trained decoder produces reconstructed frames relying on a guidance of low-level maps, without any interpolation. Experiments on a diverse set of 131 videos demonstrate that our proposed GAN-based compression engine achieves much higher quality reconstructions at very low bitrates than prevailing standard codecs such as H.264 or HEVC.
62.Noisy Computations during Inference: Harmful or Helpful? pdf
We study two aspects of noisy computations during inference. The first aspect is how to mitigate their side effects for naturally trained deep learning systems. One of the motivations for looking into this problem is to reduce the high power cost of conventional computing of neural networks through the use of analog neuromorphic circuits. Traditional GPU/CPU-centered deep learning architectures exhibit bottlenecks in power-restricted applications (e.g., embedded systems). The use of specialized neuromorphic circuits, where analog signals passed through memory-cell arrays are sensed to accomplish matrix-vector multiplications, promises large power savings and speed gains but brings with it the problems of limited precision of computations and unavoidable analog noise. We manage to improve inference accuracy from 21.1% to 99.5% for MNIST images, from 29.9% to 89.1% for CIFAR10, and from 15.5% to 89.6% for MNIST stroke sequences with the presence of strong noise (with signal-to-noise power ratio being 0 dB) by noise-injected training and a voting method. This observation promises neural networks that are insensitive to inference noise, which reduces the quality requirements on neuromorphic circuits and is crucial for their practical usage. The second aspect is how to utilize the noisy inference as a defensive architecture against black-box adversarial attacks. During inference, by injecting proper noise to signals in the neural networks, the robustness of adversarially-trained neural networks against black-box attacks has been further enhanced by 0.5% and 1.13% for two adversarially trained models for MNIST and CIFAR10, respectively.