Tsinghua-Daimler Urban Pose Dataset (coming soon)
The Tsinghua-Daimler Urban Pose Dataset (TDUP) is the first large-scale human pose estimation dataset focusing on vulnerable road users (VRUs) in urban traffic scenes. There are around 90K VRU instances manually labelled with bounding boxes, including nearly 40K pedestrians and 50K riders. The images were collected with a high-resolution on-board camera in complex Chinese urban environment.
(more)
EuroCity Persons Dataset
With over 238200 person instances manually labeled in over 47300 images, EuroCity Persons is nearly one order of magnitude larger than person datasets used previously for benchmarking. We provide full HD color images for benchmarking mono-based VRU detection. We plan to extend the dataset in the near future to provide also sequence and 3D information. Diversity is gained by recording this dataset throughout Europe.
(more)
Tsinghua-Daimler Cyclist Benchmark Dataset
The Tsinghua-Daimler Cyclist Benchmark provides a benchmark dataset for cyclist detection. We provide Full-HD color stereo images, disparity maps and vehicle data. Bounding Box based labels are provided for the classes: (“pedestrian”, “cyclist”, “motorcyclist”, “tricyclist”, “wheelchairuser”, “mopedrider”) along with various labeled tags and attributes. The dataset was recorded around Beijing area in China.
(more)
Pedestrian Segmentation Dataset
This segmentation dataset consist of contour-labeled pedestrian images captured from a vehicle-mounted calibrated stereo camera rig in an urban environment. For each pedestrian cutout we provide a 24 bit PNG image, a float disparity map and a ground truth shape.
(more)