机器学习与语料Free to access
Key facts
Contents & fields
The dataset consists of 520 scenes (260 pairs of mirrored scenarios), each 60 seconds, recording 120 frames of video and corresponding CSV rows. Folders are organized by driving mode (Baidu Apollo or manual), map (Town01-Town05), and trial number. Each trial includes:
- video——JPEG image sequences
- state_features.csv——state information of the target vehicle per frame
- detection_features.csv——2D bounding box detections from a pre-trained YOLOv7 detector
Research uses
Suitable for autonomous driving behavior analysis, human-machine driving comparison, scene understanding, object detection, etc.
This card was drafted from the source page; institution, coverage, time span, scale, fields and license are subject to the official page (pending human review).
Keywords
Why this is hard to get on your own
Obtain simulated driving scene data combining human and autonomous driving behaviors for behavior comparison and model training.