disl/my_dataset

This is the Wildtrack dataset. Annotations: Frame annotations are in the folder “annotations_positions”. Corresponding images from 7 cameras are in the folder “Image_subsets”. In order to recover the 3D location of the target, you have to use the following information. The “positionID” in .json files are indexed on a 480x1440 grid, X-first, with a 2.5cm spacing. The origin is (-3.0,-9.0) Therefore, X = -3.0 + 0.025ID%480 Y = -9.0 + 0.025ID/480 The file “rectangles.pom” gives correspondence… See the full description on the dataset page: https://huggingface.co/datasets/disl/my_dataset.