LLMOT

该数据集是一个专门为提升低光照条件下多目标跟踪技术而设计的数据集，包含了标注和未标注的行人追踪图片。其中，LLMOT数据集通过增强MOT17数据集中的夜间数据以及添加了各种场景和环境下的未标注低光照视频而构建。该数据集规模包含11,580张图片，其中有5,316张标注图片和6,264张未标注图片，其任务重点在于多目标跟踪。This dataset is specifically designed to advance multi-object tracking technology under low-light conditions, containing both annotated and unannotated pedestrian tracking images. The LLMOT dataset is constructed by enhancing the nighttime data from the MOT17 dataset and adding unannotated low-light videos across diverse scenarios and environments. In total, the dataset comprises 11,580 images, including 5,316 annotated images and 6,264 unannotated images, with its core task focusing on multi-object tracking.

搜集汇总

背景与挑战

背景概述

LLMOT数据集专门针对低光照条件下的多目标跟踪任务设计，通过增强MOT17中的夜间数据并加入多种场景的未标注低光照视频构建而成，包含11,580张图片（5,316张标注和6,264张未标注），旨在提升算法在弱光环境下的跟踪性能。