Parallel Domain Synthetic Data for the Woodscape Challenge 2023

Parallel Domain is excited to announce our partnership with Valeo for the latest iteration of the WoodScape Challenge. In previous years, Valeo has successfully hosted the competition, which has seen an overwhelming response from researchers worldwide with over 200 participants. In previous iterations, the competition focused on semantic segmentation and object detection, and the winners were given an opportunity to present their innovative solutions in a special workshop session.

Parallel Domain has developed a synthetic data platform that enables researchers to generate highly realistic, diverse, and labeled synthetic data to accelerate perception training and testing. 

This year’s challenge involves training a single model using both synthetic and real data that gets the best score on the motion segmentation task against real data (Woodscape). The annotations are at the pixel level, and the submissions will be binary masks for static/motion pixels, and the best model will have the highest mIoU on these two classes.

We look forward to seeing the innovative solutions that researchers will come up with and are excited to continue this collaboration with Valeo. The winners of the challenge will have the opportunity to present their work at our workshop special session.

To download the synthetic dataset necessary for the challenge, please click the download button in the sidebar.

Please visit the challenge webpage for more information on the competition, including details on its rules, awards, and timeline. We encourage you to participate and submit your innovative solutions to help advance computer vision for autonomous driving. Thank you for your interest in the WoodScape Challenge!