Object Detection Inference: Visualized

Object detection is a critical task in computer vision - powering use cases such as autonomous driving, surveillance, defect detection in manufacturing, medical image analysis, and more.

This project offers a blog-style Streamlit application that visually unpacks the inference operations of a modern, single-stage object detector. Specifically, we see how a RetinaNet architecture processes an image to quickly and accurately detect objects, while also exploring fundamental object detection concepts like: multi-scale feature extraction with Feature Pyramid Networks (FPNs), inline anchor box generation with Region Proposal Networks (RPNs), and detection post-processing with Non-Maximum Suppression (NMS).