Study on Tiny Object Detection

Abstract

Object detection has been one of the most challenging tasks in computer vision and a hot research topic in the world. With the rapid development of in-depth learning technology, researchers have obtained abundant research results in the field of object detection. However, most of the current mainstream object detection methods are based on the modeling of normal scale objects, and the performance of these methods is seriously degraded when they are directly applied to the detection of tiny objects, because the real scene is changing and unknown, and generally there are problems such as object occlusion, close connection and different scales. In this paper, the existing detection methods of tiny objects are summarized.