WilderPerson 户外行人检测数据集

WiderPerson 是关于户外行人检测的基准数据集。该数据集图像场景多样，不再局限于交通场景。该数据集包含 13,382 张图像，40 万个遮挡物的标注，其中 8,000 张图像用于训练，1,000 张图像用于验证，4,382 张图像用于测试。与 CityPersons 和 WIDER FACE 数据集类似，该数据集不公布测试图像的 bounding box ground truth 。该数据集包含密集的行人和各种遮挡，适合进行户外环境的行人检测评估。

WiderPerson is a benchmark dataset for outdoor pedestrian detection. The images in this dataset encompass diverse scenarios, no longer being restricted solely to traffic environments. This dataset consists of 13,382 total images with 400,000 occlusion annotations, where 8,000 images are assigned for training, 1,000 for validation, and 4,382 for testing. Similar to the CityPersons and WIDER FACE datasets, this dataset does not disclose the bounding box ground truth for its test images. Featuring dense pedestrian distributions and diverse occlusion conditions, this dataset is well-suited for evaluating pedestrian detection models in outdoor settings.

搜集汇总

背景与挑战

背景概述

WilderPerson 是一个户外行人检测基准数据集，包含13,382张多样场景的图像和40万个遮挡物标注，适用于评估密集行人和遮挡情况下的检测性能。该数据集不公布测试图像的ground truth，强调户外环境的挑战性，主要用于训练和验证行人检测模型。