1.RadarNet: Exploiting Radar for Robust Perception of Dynamic Objects ⬇️
We tackle the problem of exploiting Radar for perception in the context of self-driving as Radar provides complementary information to other sensors such as LiDAR or cameras in the form of Doppler velocity. The main challenges of using Radar are the noise and measurement ambiguities which have been a struggle for existing simple input or output fusion methods. To better address this, we propose a new solution that exploits both LiDAR and Radar sensors for perception. Our approach, dubbed RadarNet, features a voxel-based early fusion and an attention-based late fusion, which learn from data to exploit both geometric and dynamic information of Radar data. RadarNet achieves state-of-the-art results on two large-scale real-world datasets in the tasks of object detection and velocity estimation. We further show that exploiting Radar improves the perception capabilities of detecting faraway objects and understanding the motion of dynamic objects.
2.Assessing Risks of Biases in Cognitive Decision Support Systems ⬇️
Recognizing, assessing, countering, and mitigating the biases of different nature from heterogeneous sources is a critical problem in designing a cognitive Decision Support System (DSS). An example of such a system is a cognitive biometric-enabled security checkpoint. Biased algorithms affect the decision-making process in an unpredictable way, e.g. face recognition for different demographic groups may severely impact the risk assessment at a checkpoint. This paper addresses a challenging research question on how to manage an ensemble of biases? We provide performance projections of the DSS operational landscape in terms of biases. A probabilistic reasoning technique is used for assessment of the risk of such biases. We also provide a motivational experiment using face biometric component of the checkpoint system which highlights the discovery of an ensemble of biases and the techniques to assess their risks.
3.EXPO-HD: Exact Object Perception usingHigh Distraction Synthetic Data ⬇️
We present a new labeled visual dataset intended for use in object detection and segmentation tasks. This dataset consists of 5,000 synthetic photorealistic images with their corresponding pixel-perfect segmentation ground truth. The goal is to create a photorealistic 3D representation of a specific object and utilize it within a simulated training data setting to achieve high accuracy on manually gathered and annotated real-world data. Expo Markers were chosen for this task, fitting our requirements of an exact object due to the exact texture, size and 3D shape. An additional advantage is the availability of this object in offices around the world for easy testing and validation of our results. We generate the data using a domain randomization technique that also simulates other photorealistic objects in the scene, known as distraction objects. These objects provide visual complexity, occlusions, and lighting challenges to help our model gain robustness in training. We are also releasing our manually-labeled real-image test dataset. This white-paper provides strong evidence that photorealistic simulated data can be used in practical real world applications as a more scalable and flexible solution than manually-captured data. this https URL
4.Multi-level Cross-modal Interaction Network for RGB-D Salient Object Detection ⬇️
Depth cues with affluent spatial information have been proven beneficial in boosting salient object detection (SOD), while the depth quality directly affects the subsequent SOD performance. However, it is inevitable to obtain some low-quality depth cues due to limitations of its acquisition devices, which can inhibit the SOD performance. Besides, existing methods tend to combine RGB images and depth cues in a direct fusion or a simple fusion module, which makes they can not effectively exploit the complex correlations between the two sources. Moreover, few methods design an appropriate module to fully fuse multi-level features, resulting in cross-level feature interaction insufficient. To address these issues, we propose a novel Multi-level Cross-modal Interaction Network (MCINet) for RGB-D based SOD. Our MCI-Net includes two key components: 1) a cross-modal feature learning network, which is used to learn the high-level features for the RGB images and depth cues, effectively enabling the correlations between the two sources to be exploited; and 2) a multi-level interactive integration network, which integrates multi-level cross-modal features to boost the SOD performance. Extensive experiments on six benchmark datasets demonstrate the superiority of our MCI-Net over 14 state-of-the-art methods, and validate the effectiveness of different components in our MCI-Net. More important, our MCI-Net significantly improves the SOD performance as well as has a higher FPS.
5.Dive Deeper Into Box for Object Detection ⬇️
Anchor free methods have defined the new frontier in state-of-the-art object detection researches where accurate bounding box estimation is the key to the success of these methods. However, even the bounding box has the highest confidence score, it is still far from perfect at localization. To this end, we propose a box reorganization method(DDBNet), which can dive deeper into the box for more accurate localization. At the first step, drifted boxes are filtered out because the contents in these boxes are inconsistent with target semantics. Next, the selected boxes are broken into boundaries, and the well-aligned boundaries are searched and grouped into a sort of optimal boxes toward tightening instances more precisely. Experimental results show that our method is effective which leads to state-of-the-art performance for object detection.
6.On the Impact of Lossy Image and Video Compression on the Performance of Deep Convolutional Neural Network Architectures ⬇️
Recent advances in generalized image understanding have seen a surge in the use of deep convolutional neural networks (CNN) across a broad range of image-based detection, classification and prediction tasks. Whilst the reported performance of these approaches is impressive, this study investigates the hitherto unapproached question of the impact of commonplace image and video compression techniques on the performance of such deep learning architectures. Focusing on the JPEG and H.264 (MPEG-4 AVC) as a representative proxy for contemporary lossy image/video compression techniques that are in common use within network-connected image/video devices and infrastructure, we examine the impact on performance across five discrete tasks: human pose estimation, semantic segmentation, object detection, action recognition, and monocular depth estimation. As such, within this study we include a variety of network architectures and domains spanning end-to-end convolution, encoder-decoder, region-based CNN (R-CNN), dual-stream, and generative adversarial networks (GAN). Our results show a non-linear and non-uniform relationship between network performance and the level of lossy compression applied. Notably, performance decreases significantly below a JPEG quality (quantization) level of 15% and a H.264 Constant Rate Factor (CRF) of 40. However, retraining said architectures on pre-compressed imagery conversely recovers network performance by up to 78.4% in some cases. Furthermore, there is a correlation between architectures employing an encoder-decoder pipeline and those that demonstrate resilience to lossy image compression. The characteristics of the relationship between input compression to output task performance can be used to inform design decisions within future image/video devices and infrastructure.
7.Discrepancy Minimization in Domain Generalization with Generative Nearest Neighbors ⬇️
Domain generalization (DG) deals with the problem of domain shift where a machine learning model trained on multiple-source domains fail to generalize well on a target domain with different statistics. Multiple approaches have been proposed to solve the problem of domain generalization by learning domain invariant representations across the source domains that fail to guarantee generalization on the shifted target domain. We propose a Generative Nearest Neighbor based Discrepancy Minimization (GNNDM) method which provides a theoretical guarantee that is upper bounded by the error in the labeling process of the target. We employ a Domain Discrepancy Minimization Network (DDMN) that learns domain agnostic features to produce a single source domain while preserving the class labels of the data points. Features extracted from this source domain are learned using a generative model whose latent space is used as a sampler to retrieve the nearest neighbors for the target data points. The proposed method does not require access to the domain labels (a more realistic scenario) as opposed to the existing approaches. Empirically, we show the efficacy of our method on two datasets: PACS and VLCS. Through extensive experimentation, we demonstrate the effectiveness of the proposed method that outperforms several state-of-the-art DG methods.
8.Faster Mean-shift: GPU-accelerated Embedding-clustering for Cell Segmentation and Tracking ⬇️
Recently, single-stage embedding based deep learning algorithms gain increasing attention in cell segmentation and tracking. Compared with the traditional "segment-then-associate" two-stage approach, a single-stage algorithm not only simultaneously achieves consistent instance cell segmentation and tracking but also gains superior performance when distinguishing ambiguous pixels on boundaries and overlapped objects. However, the deployment of an embedding based algorithm is restricted by slow inference speed (e.g., around 1-2 mins per frame). In this study, we propose a novel Faster Mean-shift algorithm, which tackles the computational bottleneck of embedding based cell segmentation and tracking. Different from previous GPU-accelerated fast mean-shift algorithms, a new online seed optimization policy (OSOP) is introduced to adaptively determine the minimal number of seeds, accelerate computation, and save GPU memory. With both embedding simulation and empirical validation via the four cohorts from the ISBI cell tracking challenge, the proposed Faster Mean-shift algorithm achieved 7-10 times speedup compared to the state-of-the-art embedding based cell instance segmentation and tracking algorithm. Our Faster Mean-shift algorithm also achieved the highest computational speed compared to other GPU benchmarks with optimized memory consumption. The Faster Mean-shift is a plug-and-play model, which can be employed on other pixel embedding based clustering inference for medical image analysis.
9.RANDOM MASK: Towards Robust Convolutional Neural Networks ⬇️
Robustness of neural networks has recently been highlighted by the adversarial examples, i.e., inputs added with well-designed perturbations which are imperceptible to humans but can cause the network to give incorrect outputs. In this paper, we design a new CNN architecture that by itself has good robustness. We introduce a simple but powerful technique, Random Mask, to modify existing CNN structures. We show that CNN with Random Mask achieves state-of-the-art performance against black-box adversarial attacks without applying any adversarial training. We next investigate the adversarial examples which 'fool' a CNN with Random Mask. Surprisingly, we find that these adversarial examples often 'fool' humans as well. This raises fundamental questions on how to define adversarial examples and robustness properly.
10.Bayesian Multi Scale Neural Network for Crowd Counting ⬇️
Crowd Counting is a difficult but important problem in deep learning. Convolutional Neural Networks based on estimating the density map over the image has been highly successful in this domain. However dense crowd counting remains an open problem because of severe occlusion and perspective view in which people can be present at various sizes. In this work, we propose a new network which uses a ResNet based feature extractor, downsampling block using dilated convolutions and upsampling block using transposed convolutions. We present a novel aggregation module which makes our network robust to the perspective view problem. We present the optimization details, loss functions and the algorithm used in our work. On evaluating on ShanghaiTech, UCF-CC-50 and UCF-QNRF datasets using MSE and MAE as evaluation metrics, our network outperforms previous state of the art approaches while giving uncertainty estimates using a principled bayesian approach.
11.Handling confounding variables in statistical shape analysis -- application to cardiac remodelling ⬇️
Statistical shape analysis is a powerful tool to assess organ morphologies and find shape changes associated to a particular disease. However, imbalance in confounding factors, such as demographics might invalidate the analysis if not taken into consideration. Despite the methodological advances in the field, providing new methods that are able to capture complex and regional shape differences, the relationship between non-imaging information and shape variability has been overlooked. We present a linear statistical shape analysis framework that finds shape differences unassociated to a controlled set of confounding variables. It includes two confounding correction methods: confounding deflation and adjustment. We applied our framework to a cardiac magnetic resonance imaging dataset, consisting of the cardiac ventricles of 89 triathletes and 77 controls, to identify cardiac remodelling due to the practice of endurance exercise. To test robustness to confounders, subsets of this dataset were generated by randomly removing controls with low body mass index, thus introducing imbalance. The analysis of the whole dataset indicates an increase of ventricular volumes and myocardial mass in athletes, which is consistent with the clinical literature. However, when confounders are not taken into consideration no increase of myocardial mass is found. Using the downsampled datasets, we find that confounder adjustment methods are needed to find the real remodelling patterns in imbalanced datasets.
12.Structured Weight Priors for Convolutional Neural Networks ⬇️
Selection of an architectural prior well suited to a task (e.g. convolutions for image data) is crucial to the success of deep neural networks (NNs). Conversely, the weight priors within these architectures are typically left vague, e.g.~independent Gaussian distributions, which has led to debate over the utility of Bayesian deep learning. This paper explores the benefits of adding structure to weight priors. It initially considers first-layer filters of a convolutional NN, designing a prior based on random Gabor filters. Second, it considers adding structure to the prior of final-layer weights by estimating how each hidden feature relates to each class. Empirical results suggest that these structured weight priors lead to more meaningful functional priors for image data. This contributes to the ongoing discussion on the importance of weight priors.
13.A Competitive Deep Neural Network Approach for the ImageCLEFmed Caption 2020 Task ⬇️
The aim of this task is to develop a system that automatically labels radiology images with relevant medical concepts. We describe our Deep Neural Network (DNN) based approach for tackling this problem. On the challenge test set of 3,534 radiology images, our system achieves an F1 score of 0.375 and ranks high (12th among all the systems that were successfully submitted to the challenge), whereby we only rely on the provided data sources and do not use external medical knowledge or ontologies, or pretrained models from other medical image repositories or application domains.
14.Optimization of XNOR Convolution for Binary Convolutional Neural Networks on GPU ⬇️
Binary convolutional networks have lower computational load and lower memory foot-print compared to their full-precision counterparts. So, they are a feasible alternative for the deployment of computer vision applications on limited capacity embedded devices. Once trained on less resource-constrained computational environments, they can be deployed for real-time inference on such devices. In this study, we propose an implementation of binary convolutional network inference on GPU by focusing on optimization of XNOR convolution. Experimental results show that using GPU can provide a speed-up of up to
$42.61\times$ with a kernel size of$3\times3$ . The implementation is publicly available at this https URL
15.Generative networks as inverse problems with fractional wavelet scattering networks ⬇️
Deep learning is a hot research topic in the field of machine learning methods and applications. Generative Adversarial Networks (GANs) and Variational Auto-Encoders (VAEs) provide impressive image generations from Gaussian white noise, but both of them are difficult to train since they need to train the generator (or encoder) and the discriminator (or decoder) simultaneously, which is easy to cause unstable training. In order to solve or alleviate the synchronous training difficult problems of GANs and VAEs, recently, researchers propose Generative Scattering Networks (GSNs), which use wavelet scattering networks (ScatNets) as the encoder to obtain the features (or ScatNet embeddings) and convolutional neural networks (CNNs) as the decoder to generate the image. The advantage of GSNs is the parameters of ScatNets are not needed to learn, and the disadvantage of GSNs is that the expression ability of ScatNets is slightly weaker than CNNs and the dimensional reduction method of Principal Component Analysis (PCA) is easy to lead overfitting in the training of GSNs, and therefore affect the generated quality in the testing process. In order to further improve the quality of generated images while keep the advantages of GSNs, this paper proposes Generative Fractional Scattering Networks (GFRSNs), which use more expressive fractional wavelet scattering networks (FrScatNets) instead of ScatNets as the encoder to obtain the features (or FrScatNet embeddings) and use the similar CNNs of GSNs as the decoder to generate the image. Additionally, this paper develops a new dimensional reduction method named Feature-Map Fusion (FMF) instead of PCA for better keeping the information of FrScatNets and the effect of image fusion on the quality of image generation is also discussed.
16.Learning Modality Interaction for Temporal Sentence Localization and Event Captioning in Videos ⬇️
Automatically generating sentences to describe events and temporally localizing sentences in a video are two important tasks that bridge language and videos. Recent techniques leverage the multimodal nature of videos by using off-the-shelf features to represent videos, but interactions between modalities are rarely explored. Inspired by the fact that there exist cross-modal interactions in the human brain, we propose a novel method for learning pairwise modality interactions in order to better exploit complementary information for each pair of modalities in videos and thus improve performances on both tasks. We model modality interaction in both the sequence and channel levels in a pairwise fashion, and the pairwise interaction also provides some explainability for the predictions of target tasks. We demonstrate the effectiveness of our method and validate specific design choices through extensive ablation studies. Our method turns out to achieve state-of-the-art performances on four standard benchmark datasets: MSVD and MSR-VTT (event captioning task), and Charades-STA and ActivityNet Captions (temporal sentence localization task).
17.Nonnegative Low Rank Tensor Approximation and its Application to Multi-dimensional Images ⬇️
The main aim of this paper is to develop a new algorithm for computing Nonnegative Low Rank Tensor (NLRT) approximation for nonnegative tensors that arise in many multi-dimensional imaging applications. Nonnegativity is one of the important property as each pixel value refer to nonzero light intensity in image data acquisition. Our approach is different from classical nonnegative tensor factorization (NTF) which has been studied for many years. For a given nonnegative tensor, the classical NTF approach is to determine nonnegative low rank tensor by computing factor matrices or tensors (for example, CPD finds factor matrices while Tucker decomposition finds core tensor and factor matrices), such that the distance between this nonnegative low rank tensor and given tensor is as small as possible. The proposed NLRT approach is different from the classical NTF. It determines a nonnegative low rank tensor without using decompositions or factorization methods. The minimized distance by the proposed NLRT method can be smaller than that by the NTF method, and it implies that the proposed NLRT method can obtain a better low rank tensor approximation. The proposed NLRT approximation algorithm is derived by using the alternating averaged projection on the product of low rank matrix manifolds and non-negativity property. We show the convergence of the alternating projection algorithm. Experimental results for synthetic data and multi-dimensional images are presented to demonstrate the performance of the proposed NLRT method is better than that of existing NTF methods.
18.WaveFuse: A Unified Deep Framework for Image Fusion with Wavelet Transform ⬇️
We propose an unsupervised image fusion architecture for multiple application scenarios based on the combination of multi-scale discrete wavelet transform through regional energy and deep learning. To our best knowledge, this is the first time the conventional image fusion method has been combined with deep learning. The useful information of feature maps can be utilized adequately through multi-scale discrete wavelet transform in our proposed method.Compared with other state-of-the-art fusion method, the proposed algorithm exhibits better fusion performance in both subjective and objective evaluation. Moreover, it's worth mentioning that comparable fusion performance trained in COCO dataset can be obtained by training with a much smaller dataset with only hundreds of images chosen randomly from COCO. Hence, the training time is shortened substantially, leading to the improvement of the model's performance both in practicality and training efficiency.
19.A Deep Learning-based Detector for Brown SpotDisease in Passion Fruit Plant Leaves ⬇️
Pests and diseases pose a key challenge to passion fruit farmers across Uganda and East Africa in general. They lead to loss of investment as yields reduce and losses increases. As the majority of the farmers, including passion fruit farmers, in the country are smallholder farmers from low-income households, they do not have the sufficient information and means to combat these challenges. While, passion fruits have the potential to improve the well-being of these farmers as they have a short maturity period and high market value , without the required knowledge about the health of their crops, farmers cannot intervene promptly to turn the situation around.
For this work, we have partnered with the Uganda National Crop Research Institute (NaCRRI) to develop a dataset of expertly labelled passion fruit plant leaves and fruits, both diseased and healthy. We have made use of their extension service to collect images from 5 districts in Uganda,
With the dataset in place, we are employing state-of-the-art techniques in machine learning, and specifically deep learning, techniques at scale for object detection and classification to correctly determine the health status of passion fruit plants and provide an accurate diagnosis for positive detections.This work focuses on two major diseases woodiness (viral) and brown spot (fungal) diseases.
20.Toward Zero-Shot Unsupervised Image-to-Image Translation ⬇️
Recent studies have shown remarkable success in unsupervised image-to-image translation. However, if there has no access to enough images in target classes, learning a mapping from source classes to the target classes always suffers from mode collapse, which limits the application of the existing methods. In this work, we propose a zero-shot unsupervised image-to-image translation framework to address this limitation, by associating categories with their side information like attributes. To generalize the translator to previous unseen classes, we introduce two strategies for exploiting the space spanned by the semantic attributes. Specifically, we propose to preserve semantic relations to the visual space and expand attribute space by utilizing attribute vectors of unseen classes, thus encourage the translator to explore the modes of unseen classes. Quantitative and qualitative results on different datasets demonstrate the effectiveness of our proposed approach. Moreover, we demonstrate that our framework can be applied to many tasks, such as zero-shot classification and fashion design.
21.Superpixel Based Graph Laplacian Regularization for Sparse Hyperspectral Unmixing ⬇️
An efficient spatial regularization method using superpixel segmentation and graph Laplacian regularization is proposed for sparse hyperspectral unmixing method. A superpixel is defined as a group of structured neighboring pixels which constitutes a homogeneous region. First, we segment the hyperspectral image into many superpixels. Then, a weighted graph in each superpixel is constructed. Each node in the graph represents the spectrum of a pixel and edges connect the similar pixels inside the superpixel. The spatial similarity is investigated in each superpixel using graph Laplacian regularization. A weighted sparsity promoting norm is included in the formulation to sparsify the abundance matrix. Experimental results on simulated and real data sets show the superiority of the proposed algorithm over the well-known algorithms in the literature.
22.Change Detection Using Synthetic Aperture Radar Videos ⬇️
Many researches have been carried out for change detection using temporal SAR images. In this paper an algorithm for change detection using SAR videos has been proposed. There are various challenges related to SAR videos such as high level of speckle noise, rotation of SAR image frames of the video around a particular axis due to the circular movement of airborne vehicle, non-uniform back scattering of SAR pulses. Hence conventional change detection algorithms used for optical videos and SAR temporal images cannot be directly utilized for SAR videos. We propose an algorithm which is a combination of optical flow calculation using Lucas Kanade (LK) method and blob detection. The developed method follows a four steps approach: image filtering and enhancement, applying LK method, blob analysis and combining LK method with blob analysis. The performance of the developed approach was tested on SAR videos available on Sandia National Laboratories website and SAR videos generated by a SAR simulator.
23.Quantum-soft QUBO Suppression for Accurate Object Detection ⬇️
Non-maximum suppression (NMS) has been adopted by default for removing redundant object detections for decades. It eliminates false positives by only keeping the image M with highest detection score and images whose overlap ratio with M is less than a predefined threshold. However, this greedy algorithm may not work well for object detection under occlusion scenario where true positives with lower detection scores are possibly suppressed. In this paper, we first map the task of removing redundant detections into Quadratic Unconstrained Binary Optimization (QUBO) framework that consists of detection score from each bounding box and overlap ratio between pair of bounding boxes. Next, we solve the QUBO problem using the proposed Quantum-soft QUBO Suppression (QSQS) algorithm for fast and accurate detection by exploiting quantum computing advantages. Experiments indicate that QSQS improves mean average precision from 74.20% to 75.11% for PASCAL VOC 2007. It consistently outperforms NMS and soft-NMS for Reasonable subset of benchmark pedestrian detection CityPersons.
24.Monocular Real-Time Volumetric Performance Capture ⬇️
We present the first approach to volumetric performance capture and novel-view rendering at real-time speed from monocular video, eliminating the need for expensive multi-view systems or cumbersome pre-acquisition of a personalized template model. Our system reconstructs a fully textured 3D human from each frame by leveraging Pixel-Aligned Implicit Function (PIFu). While PIFu achieves high-resolution reconstruction in a memory-efficient manner, its computationally expensive inference prevents us from deploying such a system for real-time applications. To this end, we propose a novel hierarchical surface localization algorithm and a direct rendering method without explicitly extracting surface meshes. By culling unnecessary regions for evaluation in a coarse-to-fine manner, we successfully accelerate the reconstruction by two orders of magnitude from the baseline without compromising the quality. Furthermore, we introduce an Online Hard Example Mining (OHEM) technique that effectively suppresses failure modes due to the rare occurrence of challenging examples. We adaptively update the sampling probability of the training data based on the current reconstruction accuracy, which effectively alleviates reconstruction artifacts. Our experiments and evaluations demonstrate the robustness of our system to various challenging angles, illuminations, poses, and clothing styles. We also show that our approach compares favorably with the state-of-the-art monocular performance capture. Our proposed approach removes the need for multi-view studio settings and enables a consumer-accessible solution for volumetric capture.
25.Accurate, Low-Latency Visual Perception for Autonomous Racing:Challenges, Mechanisms, and Practical Solutions ⬇️
Autonomous racing provides the opportunity to test safety-critical perception pipelines at their limit. This paper describes the practical challenges and solutions to applying state-of-the-art computer vision algorithms to build a low-latency, high-accuracy perception system for DUT18 Driverless (DUT18D), a 4WD electric race car with podium finishes at all Formula Driverless competitions for which it raced. The key components of DUT18D include YOLOv3-based object detection, pose estimation, and time synchronization on its dual stereovision/monovision camera setup. We highlight modifications required to adapt perception CNNs to racing domains, improvements to loss functions used for pose estimation, and methodologies for sub-microsecond camera synchronization among other improvements. We perform a thorough experimental evaluation of the system, demonstrating its accuracy and low-latency in real-world racing scenarios.
26.Weakly Supervised 3D Object Detection from Point Clouds ⬇️
A crucial task in scene understanding is 3D object detection, which aims to detect and localize the 3D bounding boxes of objects belonging to specific classes. Existing 3D object detectors heavily rely on annotated 3D bounding boxes during training, while these annotations could be expensive to obtain and only accessible in limited scenarios. Weakly supervised learning is a promising approach to reducing the annotation requirement, but existing weakly supervised object detectors are mostly for 2D detection rather than 3D. In this work, we propose VS3D, a framework for weakly supervised 3D object detection from point clouds without using any ground truth 3D bounding box for training. First, we introduce an unsupervised 3D proposal module that generates object proposals by leveraging normalized point cloud densities. Second, we present a cross-modal knowledge distillation strategy, where a convolutional neural network learns to predict the final results from the 3D object proposals by querying a teacher network pretrained on image datasets. Comprehensive experiments on the challenging KITTI dataset demonstrate the superior performance of our VS3D in diverse evaluation settings. The source code and pretrained models are publicly available at this https URL.
27.Variants of BERT, Random Forests and SVM approach for Multimodal Emotion-Target Sub-challenge ⬇️
Emotion recognition has become a major problem in computer vision in recent years that made a lot of effort by researchers to overcome the difficulties in this task. In the field of affective computing, emotion recognition has a wide range of applications, such as healthcare, robotics, human-computer interaction. Due to its practical importance for other tasks, many techniques and approaches have been investigated for different problems and various data sources. Nevertheless, comprehensive fusion of the audio-visual and language modalities to get the benefits from them is still a problem to solve. In this paper, we present and discuss our classification methodology for MuSe-Topic Sub-challenge, as well as the data and results. For the topic classification, we ensemble two language models which are ALBERT and RoBERTa to predict 10 classes of topics. Moreover, for the classification of valence and arousal, SVM and Random forests are employed in conjunction with feature selection to enhance the performance.
28.Demystifying Contrastive Self-Supervised Learning: Invariances, Augmentations and Dataset Biases ⬇️
Self-supervised representation learning approaches have recently surpassed their supervised learning counterparts on downstream tasks like object detection and image classification. Somewhat mysteriously the recent gains in performance come from training instance classification models, treating each image and it's augmented versions as samples of a single class. In this work, we first present quantitative experiments to demystify these gains. We demonstrate that approaches like MOCO and PIRL learn occlusion-invariant representations. However, they fail to capture viewpoint and category instance invariance which are crucial components for object recognition. Second, we demonstrate that these approaches obtain further gains from access to a clean object-centric training dataset like Imagenet. Finally, we propose an approach to leverage unstructured videos to learn representations that possess higher viewpoint invariance. Our results show that the learned representations outperform MOCOv2 trained on the same data in terms of invariances encoded and the performance on downstream image classification and semantic segmentation tasks.
29.Active Learning for Video Description With Cluster-Regularized Ensemble Ranking ⬇️
Automatic video captioning aims to train models to generate text descriptions for all segments in a video, however, the most effective approaches require large amounts of manual annotation which is slow and expensive. Active learning is a promising way to efficiently build a training set for video captioning tasks while reducing the need to manually label uninformative examples. In this work we both explore various active learning approaches for automatic video captioning and show that a cluster-regularized ensemble strategy provides the best active learning approach to efficiently gather training sets for video captioning. We evaluate our approaches on the MSR-VTT and LSMDC datasets using both transformer and LSTM based captioning models and show that our novel strategy can achieve high performance while using up to 60% fewer training data than the strong state of the art baselines.
30.Deep Hashing with Hash-Consistent Large Margin Proxy Embeddings ⬇️
Image hash codes are produced by binarizing the embeddings of convolutional neural networks (CNN) trained for either classification or retrieval. While proxy embeddings achieve good performance on both tasks, they are non-trivial to binarize, due to a rotational ambiguity that encourages non-binary embeddings. The use of a fixed set of proxies (weights of the CNN classification layer) is proposed to eliminate this ambiguity, and a procedure to design proxy sets that are nearly optimal for both classification and hashing is introduced. The resulting hash-consistent large margin (HCLM) proxies are shown to encourage saturation of hashing units, thus guaranteeing a small binarization error, while producing highly discriminative hash-codes. A semantic extension (sHCLM), aimed to improve hashing performance in a transfer scenario, is also proposed. Extensive experiments show that sHCLM embeddings achieve significant improvements over state-of-the-art hashing procedures on several small and large datasets, both within and beyond the set of training classes.
31.Automatic Detection and Classification of Waste Consumer Medications for Proper Management and Disposal ⬇️
Every year, millions of pounds of medicines remain unused in the U.S. and are subject to an in-home disposal, i.e., kept in medicine cabinets, flushed in toilet or thrown in regular trash. In-home disposal, however, can negatively impact the environment and public health. The drug take-back programs (drug take-backs) sponsored by the Drug Enforcement Administration (DEA) and its state and industry partners collect unused consumer medications and provide the best alternative to in-home disposal of medicines. However, the drug take-backs are expensive to operate and not widely available. In this paper, we show that artificial intelligence (AI) can be applied to drug take-backs to render them operationally more efficient. Since identification of any waste is crucial to a proper disposal, we showed that it is possible to accurately identify loose consumer medications solely based on the physical features and visual appearance. We have developed an automatic technique that uses deep neural networks and computer vision to identify and segregate solid medicines. We applied the technique to images of about one thousand loose pills and succeeded in correctly identifying the pills with an accuracy of 0.912 and top-5 accuracy of 0.984. We also showed that hazardous pills could be distinguished from non-hazardous pills within the dataset with an accuracy of 0.984. We believe that the power of artificial intelligence could be harnessed in products that would facilitate the operation of the drug take-backs more efficiently and help them become widely available throughout the country.
32.Unsupervised Domain Adaptation in the Dissimilarity Space for Person Re-identification ⬇️
Person re-identification (ReID) remains a challenging task in many real-word video analytics and surveillance applications, even though state-of-the-art accuracy has improved considerably with the advent of deep learning (DL) models trained on large image datasets. Given the shift in distributions that typically occurs between video data captured from the source and target domains, and absence of labeled data from the target domain, it is difficult to adapt a DL model for accurate recognition of target data. We argue that for pair-wise matchers that rely on metric learning, e.g., Siamese networks for person ReID, the unsupervised domain adaptation (UDA) objective should consist in aligning pair-wise dissimilarity between domains, rather than aligning feature representations. Moreover, dissimilarity representations are more suitable for designing open-set ReID systems, where identities differ in the source and target domains. In this paper, we propose a novel Dissimilarity-based Maximum Mean Discrepancy (D-MMD) loss for aligning pair-wise distances that can be optimized via gradient descent. From a person ReID perspective, the evaluation of D-MMD loss is straightforward since the tracklet information allows to label a distance vector as being either within-class or between-class. This allows approximating the underlying distribution of target pair-wise distances for D-MMD loss optimization, and accordingly align source and target distance distributions. Empirical results with three challenging benchmark datasets show that the proposed D-MMD loss decreases as source and domain distributions become more similar. Extensive experimental evaluation also indicates that UDA methods that rely on the D-MMD loss can significantly outperform baseline and state-of-the-art UDA methods for person ReID without the common requirement for data augmentation and/or complex networks.
33.3DMaterialGAN: Learning 3D Shape Representation from Latent Space for Materials Science Applications ⬇️
In the field of computer vision, unsupervised learning for 2D object generation has advanced rapidly in the past few years. However, 3D object generation has not garnered the same attention or success as its predecessor. To facilitate novel progress at the intersection of computer vision and materials science, we propose a 3DMaterialGAN network that is capable of recognizing and synthesizing individual grains whose morphology conforms to a given 3D polycrystalline material microstructure. This Generative Adversarial Network (GAN) architecture yields complex 3D objects from probabilistic latent space vectors with no additional information from 2D rendered images. We show that this method performs comparably or better than state-of-the-art on benchmark annotated 3D datasets, while also being able to distinguish and generate objects that are not easily annotated, such as grain morphologies. The value of our algorithm is demonstrated with analysis on experimental real-world data, namely generating 3D grain structures found in a commercially relevant wrought titanium alloy, which were validated through statistical shape comparison. This framework lays the foundation for the recognition and synthesis of polycrystalline material microstructures, which are used in additive manufacturing, aerospace, and structural design applications.
34.Perpetual Motion: Generating Unbounded Human Motion ⬇️
The modeling of human motion using machine learning methods has been widely studied. In essence it is a time-series modeling problem involving predicting how a person will move in the future given how they moved in the past. Existing methods, however, typically have a short time horizon, predicting a only few frames to a few seconds of human motion. Here we focus on long-term prediction; that is, generating long sequences (potentially infinite) of human motion that is plausible. Furthermore, we do not rely on a long sequence of input motion for conditioning, but rather, can predict how someone will move from as little as a single pose. Such a model has many uses in graphics (video games and crowd animation) and vision (as a prior for human motion estimation or for dataset creation). To address this problem, we propose a model to generate non-deterministic, \textit{ever-changing}, perpetual human motion, in which the global trajectory and the body pose are cross-conditioned. We introduce a novel KL-divergence term with an implicit, unknown, prior. We train this using a heavy-tailed function of the KL divergence of a white-noise Gaussian process, allowing latent sequence temporal dependency. We perform systematic experiments to verify its effectiveness and find that it is superior to baseline methods.
35.A Unified Framework of Surrogate Loss by Refactoring and Interpolation ⬇️
We introduce UniLoss, a unified framework to generate surrogate losses for training deep networks with gradient descent, reducing the amount of manual design of task-specific surrogate losses. Our key observation is that in many cases, evaluating a model with a performance metric on a batch of examples can be refactored into four steps: from input to real-valued scores, from scores to comparisons of pairs of scores, from comparisons to binary variables, and from binary variables to the final performance metric. Using this refactoring we generate differentiable approximations for each non-differentiable step through interpolation. Using UniLoss, we can optimize for different tasks and metrics using one unified framework, achieving comparable performance compared with task-specific losses. We validate the effectiveness of UniLoss on three tasks and four datasets. Code is available at this https URL.
36.Robust Image Retrieval-based Visual Localization using Kapture ⬇️
In this paper, we present a versatile method for visual localization. It is based on robust image retrieval for coarse camera pose estimation and robust local features for accurate pose refinement. Our method is top ranked on various public datasets showing its ability of generalization and its great variety of applications. To facilitate experiments, we introduce kapture, a flexible data format and processing pipeline for structure from motion and visual localization that is released open source. We furthermore provide all datasets used in this paper in the kapture format to facilitate research and data processing. The code can be found at this https URL, the datasets as well as more information, updates, and news can be found at this https URL.
37.se(3)-TrackNet: Data-driven 6D Pose Tracking by Calibrating Image Residuals in Synthetic Domains ⬇️
Tracking the 6D pose of objects in video sequences is important for robot manipulation. This task, however, introduces multiple challenges: (i) robot manipulation involves significant occlusions; (ii) data and annotations are troublesome and difficult to collect for 6D poses, which complicates machine learning solutions, and (iii) incremental error drift often accumulates in long term tracking to necessitate re-initialization of the object's pose. This work proposes a data-driven optimization approach for long-term, 6D pose tracking. It aims to identify the optimal relative pose given the current RGB-D observation and a synthetic image conditioned on the previous best estimate and the object's model. The key contribution in this context is a novel neural network architecture, which appropriately disentangles the feature encoding to help reduce domain shift, and an effective 3D orientation representation via Lie Algebra. Consequently, even when the network is trained only with synthetic data can work effectively over real images. Comprehensive experiments over benchmarks - existing ones as well as a new dataset with significant occlusions related to object manipulation - show that the proposed approach achieves consistently robust estimates and outperforms alternatives, even though they have been trained with real images. The approach is also the most computationally efficient among the alternatives and achieves a tracking frequency of 90.9Hz.
38.Saliency Prediction with External Knowledge ⬇️
The last decades have seen great progress in saliency prediction, with the success of deep neural networks that are able to encode high-level semantics. Yet, while humans have the innate capability in leveraging their knowledge to decide where to look (e.g. people pay more attention to familiar faces such as celebrities), saliency prediction models have only been trained with large eye-tracking datasets. This work proposes to bridge this gap by explicitly incorporating external knowledge for saliency models as humans do. We develop networks that learn to highlight regions by incorporating prior knowledge of semantic relationships, be it general or domain-specific, depending on the task of interest. At the core of the method is a new Graph Semantic Saliency Network (GraSSNet) that constructs a graph that encodes semantic relationships learned from external knowledge. A Spatial Graph Attention Network is then developed to update saliency features based on the learned graph. Experiments show that the proposed model learns to predict saliency from the external knowledge and outperforms the state-of-the-art on four saliency benchmarks.
39.Adaptive LiDAR Sampling and Depth Completion using Ensemble Variance ⬇️
This work considers the problem of depth completion, with or without image data, where an algorithm may measure the depth of a prescribed limited number of pixels. The algorithmic challenge is to choose pixel positions strategically and dynamically to maximally reduce overall depth estimation error. This setting is realized in daytime or nighttime depth completion for autonomous vehicles with a programmable LiDAR. Our method uses an ensemble of predictors to define a sampling probability over pixels. This probability is proportional to the variance of the predictions of ensemble members, thus highlighting pixels that are difficult to predict. By additionally proceeding in several prediction phases, we effectively reduce redundant sampling of similar pixels. Our ensemble-based method may be implemented using any depth-completion learning algorithm, such as a state-of-the-art neural network, treated as a black box. In particular, we also present a simple and effective Random Forest-based algorithm, and similarly use its internal ensemble in our design. We conduct experiments on the KITTI dataset, using the neural network algorithm of Ma et al. and our Random Forest based learner for implementing our method. The accuracy of both implementations exceeds the state of the art. Compared with a random or grid sampling pattern, our method allows a reduction by a factor of 4-10 in the number of measurements required to attain the same accuracy.
40.Chest X-ray Report Generation through Fine-Grained Label Learning ⬇️
Obtaining automated preliminary read reports for common exams such as chest X-rays will expedite clinical workflows and improve operational efficiencies in hospitals. However, the quality of reports generated by current automated approaches is not yet clinically acceptable as they cannot ensure the correct detection of a broad spectrum of radiographic findings nor describe them accurately in terms of laterality, anatomical location, severity, etc. In this work, we present a domain-aware automatic chest X-ray radiology report generation algorithm that learns fine-grained description of findings from images and uses their pattern of occurrences to retrieve and customize similar reports from a large report database. We also develop an automatic labeling algorithm for assigning such descriptors to images and build a novel deep learning network that recognizes both coarse and fine-grained descriptions of findings. The resulting report generation algorithm significantly outperforms the state of the art using established score metrics.
41.Corner Proposal Network for Anchor-free, Two-stage Object Detection ⬇️
The goal of object detection is to determine the class and location of objects in an image. This paper proposes a novel anchor-free, two-stage framework which first extracts a number of object proposals by finding potential corner keypoint combinations and then assigns a class label to each proposal by a standalone classification stage. We demonstrate that these two stages are effective solutions for improving recall and precision, respectively, and they can be integrated into an end-to-end network. Our approach, dubbed Corner Proposal Network (CPN), enjoys the ability to detect objects of various scales and also avoids being confused by a large number of false-positive proposals. On the MS-COCO dataset, CPN achieves an AP of 49.2% which is competitive among state-of-the-art object detection methods. CPN also fits the scenario of computational efficiency, which achieves an AP of 41.6%/39.7% at 26.2/43.3 FPS, surpassing most competitors with the same inference speed. Code is available at this https URL
42.Flower: A Friendly Federated Learning Research Framework ⬇️
Federated Learning (FL) has emerged as a promising technique for edge devices to collaboratively learn a shared prediction model, while keeping their training data on the device, thereby decoupling the ability to do machine learning from the need to store the data in the cloud. However, FL is difficult to implement and deploy in practice, considering the heterogeneity in mobile devices, e.g., different programming languages, frameworks, and hardware accelerators. Although there are a few frameworks available to simulate FL algorithms (e.g., TensorFlow Federated), they do not support implementing FL workloads on mobile devices. Furthermore, these frameworks are designed to simulate FL in a server environment and hence do not allow experimentation in distributed mobile settings for a large number of clients. In this paper, we present Flower (https://flower.dev/), a FL framework which is both agnostic towards heterogeneous client environments and also scales to a large number of clients, including mobile and embedded devices. Flower's abstractions let developers port existing mobile workloads with little overhead, regardless of the programming language or ML framework used, while also allowing researchers flexibility to experiment with novel approaches to advance the state-of-the-art. We describe the design goals and implementation considerations of Flower and show our experiences in evaluating the performance of FL across clients with heterogeneous computational and communication capabilities.
43.Monochrome and Color Polarization Demosaicking Using Edge-Aware Residual Interpolation ⬇️
A division-of-focal-plane or microgrid image polarimeter enables us to acquire a set of polarization images in one shot. Since the polarimeter consists of an image sensor equipped with a monochrome or color polarization filter array (MPFA or CPFA), the demosaicking process to interpolate missing pixel values plays a crucial role in obtaining high-quality polarization images. In this paper, we propose a novel MPFA demosaicking method based on edge-aware residual interpolation (EARI) and also extend it to CPFA demosaicking. The key of EARI is a new edge detector for generating an effective guide image used to interpolate the missing pixel values. We also present a newly constructed full color-polarization image dataset captured using a 3-CCD camera and a rotating polarizer. Using the dataset, we experimentally demonstrate that our EARI-based method outperforms existing methods in MPFA and CPFA demosaicking.
44.Efficient adaptation of neural network filter for video compression ⬇️
We present an efficient finetuning methodology for neural-network filters which are applied as a postprocessing artifact-removal step in video coding pipelines. The fine-tuning is performed at encoder side to adapt the neural network to the specific content that is being encoded. In order to maximize the PSNR gain and minimize the bitrate overhead, we propose to finetune only the convolutional layers' biases. The proposed method achieves convergence much faster than conventional finetuning approaches, making it suitable for practical applications. The weight-update can be included into the video bitstream generated by the existing video codecs. We show that our method achieves up to 9.7% average BD-rate gain when compared to the state-of-art Versatile Video Coding (VVC) standard codec on 7 test sequences.
45.Multi-camera Torso Pose Estimation using Graph Neural Networks ⬇️
Estimating the location and orientation of humans is an essential skill for service and assistive robots. To achieve a reliable estimation in a wide area such as an apartment, multiple RGBD cameras are frequently used. Firstly, these setups are relatively expensive. Secondly, they seldom perform an effective data fusion using the multiple camera sources at an early stage of the processing pipeline. Occlusions and partial views make this second point very relevant in these scenarios. The proposal presented in this paper makes use of graph neural networks to merge the information acquired from multiple camera sources, achieving a mean absolute error below 125 mm for the location and 10 degrees for the orientation using low-resolution RGB images. The experiments, conducted in an apartment with three cameras, benchmarked two different graph neural network implementations and a third architecture based on fully connected layers. The software used has been released as open-source in a public repository (this https URL).
46.Reachable Sets of Classifiers & Regression Models: (Non-)Robustness Analysis and Robust Training ⬇️
Neural networks achieve outstanding accuracy in classification and regression tasks. However, understanding their behavior still remains an open challenge that requires questions to be addressed on the robustness, explainability and reliability of predictions. We answer these questions by computing reachable sets of neural networks, i.e. sets of outputs resulting from continuous sets of inputs. We provide two efficient approaches that lead to over- and under-approximations of the reachable set. This principle is highly versatile, as we show. First, we analyze and enhance the robustness properties of both classifiers and regression models. This is in contrast to existing works, which only handle classification. Specifically, we verify (non-)robustness, propose a robust training procedure, and show that our approach outperforms adversarial attacks as well as state-of-the-art methods of verifying classifiers for non-norm bound perturbations. We also provide a technique of distinguishing between reliable and non-reliable predictions for unlabeled inputs, quantify the influence of each feature on a prediction, and compute a feature ranking.
47.DeScarGAN: Disease-Specific Anomaly Detection with Weak Supervision ⬇️
Anomaly detection and localization in medical images is a challenging task, especially when the anomaly exhibits a change of existing structures, e.g., brain atrophy or changes in the pleural space due to pleural effusions. In this work, we present a weakly supervised and detail-preserving method that is able to detect structural changes of existing anatomical structures. In contrast to standard anomaly detection methods, our method extracts information about the disease characteristics from two groups: a group of patients affected by the same disease and a healthy control group. Together with identity-preserving mechanisms, this enables our method to extract highly disease-specific characteristics for a more detailed detection of structural changes. We designed a specific synthetic data set to evaluate and compare our method against state-of-the-art anomaly detection methods. Finally, we show the performance of our method on chest X-ray images. Our method called DeScarGAN outperforms other anomaly detection methods on the synthetic data set and by visual inspection on the chest X-ray image data set.
48.Risk-Averse MPC via Visual-Inertial Input and Recurrent Networks for Online Collision Avoidance ⬇️
In this paper, we propose an online path planning architecture that extends the model predictive control (MPC) formulation to consider future location uncertainties for safer navigation through cluttered environments. Our algorithm combines an object detection pipeline with a recurrent neural network (RNN) which infers the covariance of state estimates through each step of our MPC's finite time horizon. The RNN model is trained on a dataset that comprises of robot and landmark poses generated from camera images and inertial measurement unit (IMU) readings via a state-of-the-art visual-inertial odometry framework. To detect and extract object locations for avoidance, we use a custom-trained convolutional neural network model in conjunction with a feature extractor to retrieve 3D centroid and radii boundaries of nearby obstacles. The robustness of our methods is validated on complex quadruped robot dynamics and can be generally applied to most robotic platforms, demonstrating autonomous behaviors that can plan fast and collision-free paths towards a goal point.
49.Coupled Convolutional Neural Network with Adaptive Response Function Learning for Unsupervised Hyperspectral Super-Resolution ⬇️
Due to the limitations of hyperspectral imaging systems, hyperspectral imagery (HSI) often suffers from poor spatial resolution, thus hampering many applications of the imagery. Hyperspectral super-resolution refers to fusing HSI and MSI to generate an image with both high spatial and high spectral resolutions. Recently, several new methods have been proposed to solve this fusion problem, and most of these methods assume that the prior information of the Point Spread Function (PSF) and Spectral Response Function (SRF) are known. However, in practice, this information is often limited or unavailable. In this work, an unsupervised deep learning-based fusion method - HyCoNet - that can solve the problems in HSI-MSI fusion without the prior PSF and SRF information is proposed. HyCoNet consists of three coupled autoencoder nets in which the HSI and MSI are unmixed into endmembers and abundances based on the linear unmixing model. Two special convolutional layers are designed to act as a bridge that coordinates with the three autoencoder nets, and the PSF and SRF parameters are learned adaptively in the two convolution layers during the training process. Furthermore, driven by the joint loss function, the proposed method is straightforward and easily implemented in an end-to-end training manner. The experiments performed in the study demonstrate that the proposed method performs well and produces robust results for different datasets and arbitrary PSFs and SRFs.
50.Spectral Superresolution of Multispectral Imagery with Joint Sparse and Low-Rank Learning ⬇️
Extensive attention has been widely paid to enhance the spatial resolution of hyperspectral (HS) images with the aid of multispectral (MS) images in remote sensing. However, the ability in the fusion of HS and MS images remains to be improved, particularly in large-scale scenes, due to the limited acquisition of HS images. Alternatively, we super-resolve MS images in the spectral domain by the means of partially overlapped HS images, yielding a novel and promising topic: spectral superresolution (SSR) of MS imagery. This is challenging and less investigated task due to its high ill-posedness in inverse imaging. To this end, we develop a simple but effective method, called joint sparse and low-rank learning (J-SLoL), to spectrally enhance MS images by jointly learning low-rank HS-MS dictionary pairs from overlapped regions. J-SLoL infers and recovers the unknown hyperspectral signals over a larger coverage by sparse coding on the learned dictionary pair. Furthermore, we validate the SSR performance on three HS-MS datasets (two for classification and one for unmixing) in terms of reconstruction, classification, and unmixing by comparing with several existing state-of-the-art baselines, showing the effectiveness and superiority of the proposed J-SLoL algorithm. Furthermore, the codes and datasets will be available at: this https URL_TGRS_J-SLoL, contributing to the RS community.
51.Robust Ego and Object 6-DoF Motion Estimation and Tracking ⬇️
The problem of tracking self-motion as well as motion of objects in the scene using information from a camera is known as multi-body visual odometry and is a challenging task. This paper proposes a robust solution to achieve accurate estimation and consistent track-ability for dynamic multi-body visual odometry. A compact and effective framework is proposed leveraging recent advances in semantic instance-level segmentation and accurate optical flow estimation. A novel formulation, jointly optimizing SE(3) motion and optical flow is introduced that improves the quality of the tracked points and the motion estimation accuracy. The proposed approach is evaluated on the virtual KITTI Dataset and tested on the real KITTI Dataset, demonstrating its applicability to autonomous driving applications. For the benefit of the community, we make the source code public.
52.Self-supervised Neural Audio-Visual Sound Source Localization via Probabilistic Spatial Modeling ⬇️
Detecting sound source objects within visual observation is important for autonomous robots to comprehend surrounding environments. Since sounding objects have a large variety with different appearances in our living environments, labeling all sounding objects is impossible in practice. This calls for self-supervised learning which does not require manual labeling. Most of conventional self-supervised learning uses monaural audio signals and images and cannot distinguish sound source objects having similar appearances due to poor spatial information in audio signals. To solve this problem, this paper presents a self-supervised training method using 360° images and multichannel audio signals. By incorporating with the spatial information in multichannel audio signals, our method trains deep neural networks (DNNs) to distinguish multiple sound source objects. Our system for localizing sound source objects in the image is composed of audio and visual DNNs. The visual DNN is trained to localize sound source candidates within an input image. The audio DNN verifies whether each candidate actually produces sound or not. These DNNs are jointly trained in a self-supervised manner based on a probabilistic spatial audio model. Experimental results with simulated data showed that the DNNs trained by our method localized multiple speakers. We also demonstrate that the visual DNN detected objects including talking visitors and specific exhibits from real data recorded in a science museum.
53.KOVIS: Keypoint-based Visual Servoing with Zero-Shot Sim-to-Real Transfer for Robotics Manipulation ⬇️
We present KOVIS, a novel learning-based, calibration-free visual servoing method for fine robotic manipulation tasks with eye-in-hand stereo camera system. We train the deep neural network only in the simulated environment; and the trained model could be directly used for real-world visual servoing tasks. KOVIS consists of two networks. The first keypoint network learns the keypoint representation from the image using with an autoencoder. Then the visual servoing network learns the motion based on keypoints extracted from the camera image. The two networks are trained end-to-end in the simulated environment by self-supervised learning without manual data labeling. After training with data augmentation, domain randomization, and adversarial examples, we are able to achieve zero-shot sim-to-real transfer to real-world robotic manipulation tasks. We demonstrate the effectiveness of the proposed method in both simulated environment and real-world experiment with different robotic manipulation tasks, including grasping, peg-in-hole insertion with 4mm clearance, and M13 screw insertion. The demo video is available at this http URL
54.EasierPath: An Open-source Tool for Human-in-the-loop Deep Learning of Renal Pathology ⬇️
Considerable morphological phenotyping studies in nephrology have emerged in the past few years, aiming to discover hidden regularities between clinical and imaging phenotypes. Such studies have been largely enabled by deep learning based image analysis to extract sparsely located targeting objects (e.g., glomeruli) on high-resolution whole slide images (WSI). However, such methods need to be trained using labor-intensive high-quality annotations, ideally labeled by pathologists. Inspired by the recent "human-in-the-loop" strategy, we developed EasierPath, an open-source tool to integrate human physicians and deep learning algorithms for efficient large-scale pathological image quantification as a loop. Using EasierPath, physicians are able to (1) optimize the recall and precision of deep learning object detection outcomes adaptively, (2) seamlessly support deep learning outcomes refining using either our EasierPath or prevalent ImageScope software without changing physician's user habit, and (3) manage and phenotype each object with user-defined classes. As a user case of EasierPath, we present the procedure of curating large-scale glomeruli in an efficient human-in-the-loop fashion (with two loops). From the experiments, the EasierPath saved 57 % of the annotation efforts to curate 8,833 glomeruli during the second loop. Meanwhile, the average precision of glomerular detection was leveraged from 0.504 to 0.620. The EasierPath software has been released as open-source to enable the large-scale glomerular prototyping. The code can be found in this https URL
55.Improving Lesion Segmentation for Diabetic Retinopathy using Adversarial Learning ⬇️
Diabetic Retinopathy (DR) is a leading cause of blindness in working age adults. DR lesions can be challenging to identify in fundus images, and automatic DR detection systems can offer strong clinical value. Of the publicly available labeled datasets for DR, the Indian Diabetic Retinopathy Image Dataset (IDRiD) presents retinal fundus images with pixel-level annotations of four distinct lesions: microaneurysms, hemorrhages, soft exudates and hard exudates. We utilize the HEDNet edge detector to solve a semantic segmentation task on this dataset, and then propose an end-to-end system for pixel-level segmentation of DR lesions by incorporating HEDNet into a Conditional Generative Adversarial Network (cGAN). We design a loss function that adds adversarial loss to segmentation loss. Our experiments show that the addition of the adversarial loss improves the lesion segmentation performance over the baseline.
56.Learned Pre-Processing for Automatic Diabetic Retinopathy Detection on Eye Fundus Images ⬇️
Diabetic Retinopathy is the leading cause of blindness in the working-age population of the world. The main aim of this paper is to improve the accuracy of Diabetic Retinopathy detection by implementing a shadow removal and color correction step as a preprocessing stage from eye fundus images. For this, we rely on recent findings indicating that application of image dehazing on the inverted intensity domain amounts to illumination compensation. Inspired by this work, we propose a Shadow Removal Layer that allows us to learn the pre-processing function for a particular task. We show that learning the pre-processing function improves the performance of the network on the Diabetic Retinopathy detection task.