文献类型：期刊文献

英文题名：A Review of Small Object Detection Based on Deep Learning

作者：Peng, Xiaohong[1]; Jiang, Hongbin[1]

机构：[1] College of Mathematics and Computer Science, Guangdong Ocean University, Guangdong, Zhanjiang, China

年份：2026

起止页码：89

外文期刊名：BDAIA 2025 - 2025 2nd International Conference on Big Data Analytics and Artificial Intelligence Application

收录：EI(收录号：20261020204168)

语种：英文

外文关键词：Object detection - Object recognition

外文摘要：In recent years, the technology of small object detection has continued to develop, resulting in numerous datasets and methods. This paper summarizes the major advancements in small object detection technology over the past decade, and analyzes 18 classic small object detection datasets. Secondly, this paper briefly describes small object detection methods, focusing on three typical methods to analyze their basic principles. Finally, this paper analyzes and forecasts the future development direction of small target detection. Future focus for small object detection should be on model architecture, attention mechanisms, multi-scale feature hierarchy, adaptive learning, and lightweight design to improve its performance and applicability. ? 2025 Copyright held by the owner/author(s).