Applications of Deep Learning in Object Detection
Ke Li · 2022 International Conference on Computers, Information Processing and Advanced Education (CIPAE) · 2022
Computer vision includes many branches, among which image classification, object detection, image segmentation, and object tracking are the most important research topics in the field of computer vision. Object detection is the foundation for tackling more difficult and higher-level visual tasks, including segmentation, scene understanding, object tracking, image description, event detection, and activity recognition, since it is the baseline of image understanding and computer vision. In the past decade, deep learning object detection algorithms based on CNN have developed rapidly, and many detection algorithms have emerged with better performance than traditional or earlier algorithms. This paper summarizes the representative object detection algorithms proposed in recent years and introduces them in two parts: region proposal-based methods and regression/classification-based methods. By introducing the idea in detail and the main features of each algorithm and the network structure, it helps one understand the two methods and their representative algorithms and the existing problems in the object detection field.