Research on Keypoint-based Object Detection

With the help of deep learning, especially the development of convolutional neural networks, object detection has been greatly promoted and developed. Keypoint-based object detection approach uses keypoints for objection detection and is a new research focus. This paper sorts out the development of keypoint-based object detection, analyzes and compares representative networks, and summarizes loss functions in training procedures, and discusses the problems and development of keypoint-based object detection trends.