Object Detection Part 4: Fast Detection Models
Lilian Weng
This article reviews fast object detection models including YOLO, SSD, and YOLOv2, contrasting one-stage detectors that skip region proposals with two-stage detectors like R-CNN. YOLO (2016) splits images into S×S grids and predicts bounding boxes directly, while SSD uses pyramidal feature hierarchies with 6 anchor boxes per location across multiple scales. YOLOv2 represents an enhanced version of the original YOLO framework for real-time detection applications.
Why it matters
In Part 3, we have reviewed models in the R-CNN family. All of them are region-based object detection algorithms. They can achieve high accuracy but could be too slow for certain applications such as autonomous driving. In Part 4, we only focus on fast object detection models, including SSD, RetinaNet, and models in the YOLO family.
