Object365通用物体检测数据集

数据集依托项目课题三《图像感知国家新一代人工智能开放创新平台服务机制研究和建设》建设。课题面向图像感知领域的开放平台建设，本数据集为多元开放主体提供数据开源开放服务，面向图像感知中的通用物体检测研究，提供底层数据支持。数据集由来自互联网，经过三段式标注与质检，包含超过200万张图片、365个类比、超过2850万个检测框，数据量达到380GB。

This dataset is developed under Project Topic 3: "Research and Construction of Service Mechanism for the National New-generation Artificial Intelligence Open Innovation Platform for Image Perception". The topic focuses on the construction of open platforms in the domain of image perception. This dataset provides open-source data services for various open stakeholders, and offers underlying data support for general object detection research in the field of image perception. Sourced from the Internet, the dataset has undergone three-stage annotation and quality inspection, containing over 2 million images, 365 categories, more than 28.5 million detection bounding boxes, with a total data volume of 380 GB.

搜集汇总

背景与挑战

背景概述

Object365通用物体检测数据集是一个面向图像感知领域的开放数据集，包含超过200万张图片、365个类别和2850万个检测框，总数据量达380GB。该数据集为通用物体检测研究提供底层数据支持，数据格式包括gz、json和docx。