Analysis of Anchor-Based and Anchor-Free Object Detection Methods Based on Deep Learning
Shujian Liu, Haibo Zhou, Chenming Li, Shuo Wang
Abstract
As one of the core mission of computer vision, object detection has been widely applied with the rapid development of computer technology, especially in the fields of face detection, behavior detection, auto driving, and intelligent monitoring. Aiming at the shortcomings of the traditional object detection, such as low detection accuracy, low efficiency and poor robustness, this article summarizes deep learning-based detectors of two kinds of modules: Anchor-Based and Anchor-Free. The performance of each detector is compared and analyzed in this paper as well. In addition, we summarize the development of object detection's key technologies from the aspect of improvement of Backbone, the optimization of NMS, imbalance of positive and negative samples' solution etc. Finally, the development trend of object detection is discussed from the prospect of lightweight detection model, weakly supervision detection and small object detection etc.