大规模RGB-热红外域增量目标检测数据集

数据集名称：大规模RGB-热红外域增量目标检测数据集（编号2020AAA0108902-003）数据集内容：建立了一个全新的大规模RGB-热红外（RGB-Thermal）域增量目标检测（RTDOD）数据集，旨在弥补现有数据集的不足并满足实际应用的需求。RTDOD数据集采用校准后的彩色和热像仪器，同步收集RGB图像和热红外图像，涵盖多种不同的场景。数据集总计标注了约16,200对图像，包括目标的边界框和类别信息。该数据集不仅为机器人视觉理解领域提供了宝贵的资源，还为域增量学习等新兴研究领域设定了新的测试基准。数据来源：通过现场采集获得。采集时间及地点：数据采集时间为2021年，地点为中国科学院自动化研究所。本数据集名为RTDOD，旨在弥补现有RGB-Thermal域增量目标检测数据集的不足，并满足实际应用需求。数据集采用校准后的彩色和热像仪器同步收集RGB图像和热红外图像，覆盖多种不同场景。数据量为30GB。

Dataset Name: Large-Scale RGB-Thermal Domain Incremental Object Detection Dataset (No. 2020AAA0108902-003)
Dataset Content: This work develops a brand-new large-scale RGB-Thermal Domain Incremental Object Detection (RTDOD) dataset, which aims to compensate for the limitations of existing datasets and meet the demands of practical applications. The RTDOD dataset utilizes calibrated color and thermal imaging instruments to synchronously collect RGB images and thermal infrared images across a wide range of distinct scenarios. In total, approximately 16,200 pairs of images are annotated, with bounding boxes and category information for the targets provided. This dataset not only provides a valuable resource for the field of robotic visual understanding, but also sets a new benchmark for emerging research domains such as domain incremental learning.
Data Source: Acquired through on-site collection.
Collection Time and Location: The data was collected in 2021 at the Institute of Automation, Chinese Academy of Sciences.
This dataset, named RTDOD, is designed to address the shortcomings of existing RGB-Thermal domain incremental object detection datasets and fulfill practical application requirements. It employs calibrated color and thermal imaging devices to synchronously gather RGB and thermal infrared images, covering multiple diverse scenarios. The total data volume amounts to 30 GB.

搜集汇总

背景与挑战

背景概述

该数据集是一个大规模RGB-热红外域增量目标检测数据集（RTDOD），包含约16,200对同步采集的RGB和热红外图像，覆盖多种场景并标注了目标边界框和类别信息，总数据量约30GB。它旨在弥补现有数据集的不足，为机器人视觉理解和域增量学习等领域提供资源与测试基准。数据来源于2021年在中国科学院自动化研究所的现场采集，支持国家重点研发计划项目。