目标入侵图片数据集

项目研究和实施过程中形成的目标入侵图片数据集，包括不同人员、不用光照、不同背景下的目标彩色图，用于支持目标检测算法训练与模型优化。人员图片经过分类、筛选、数据增广、标签标注后，数据集最终构成包含不同人员信息以及不同环境下的目标图像库该数据来自于2019-2022年间课题单位的试验现场，采集设备为三目相机

This target intrusion image dataset was developed during the research and implementation phases of the project. It comprises target color images captured under diverse personnel configurations, lighting conditions and background scenarios, and is designed to support target detection algorithm training and model optimization. The personnel-involved target images have been processed through classification, screening, data augmentation and label annotation, and the finalized dataset constitutes a target image library encompassing diverse personnel information and target images collected across various environmental conditions. This dataset was gathered from the test sites of the project's affiliated institution between 2019 and 2022, utilizing a trinocular camera as the data acquisition device.

搜集汇总

背景与挑战

背景概述

目标入侵图片数据集是一个包含不同人员、光照和背景下的目标彩色图的数据集，用于支持目标检测算法训练与模型优化。数据集由苏州大学陈国栋发布，包含289个文件，总数据量为31.24MB，采集于2019-2022年间的试验现场。