WilderPerson Outdoor Pedestrian Detection Dataset

WilderPerson is a benchmark dataset for outdoor pedestrian detection. The image scenes of this dataset are diverse and are no longer limited to traffic scenes. The dataset contains 13,382 images and 400,000 occluded object annotations, of which 8,000 images are used for training, 1,000 images are used for verification, and 4,382 images are used for testing. CityPersons and WIDER FACE DatasetSimilarly, this dataset does not publish the bounding box ground truth of the test image. This dataset contains dense pedestrians and various occlusions, which is suitable for pedestrian detection evaluation in outdoor environments.