Introduction
This wiki document contains information about the different libraries/frameworks that we are using in order to achieve custom object detection in Java.
- YOLO is an abbreviation for the term ‘You Only Look Once’.
- It is an algorithm that uses neural networks to provide real-time object detection.
- It employs convolutional neural networks (CNN) to detect objects in real-time.
- It has many version eg. yolo, yolo2, yolo3, tiny-Yolo.
- We are using YOLO v5 which is faster and light weight as compared to its previous version.
- It supports Google collab environment in order to train YOLO Model using google cloud GPUs.
- Deep Java Library (DJL) is an open-source, high-level, engine-agnostic Java framework for deep learning.
- You can use your existing Java expertise as an on-ramp to learn and use machine learning and deep learning.
- You can make your own models and train them using DJL.
- Or you can integrate your Models created with TensorFlow, PyTorch, ONNX, Apache MXNet into the Java application.
- DJL makes it easy to integrate your java application with DL Models.
- We are using this online software to annotate and label our images to create our dataset.
- Roboflow provides lot of functionalities to our dataset like - Annotation, Augmentation, Exporting.
- Here we are using Roboflow in order to augment our dataset.