Checkpoints for Challgenge "CVPPA@ICCV'23: Hierarchical Panoptic Segmentation of Crops and Weeds" (3rd rank)

For the task of "Hierarchical Panoptic Segmentation of Crops and Weeds", we trained three different Mask2Former models on the different granularity levels, leading to a final score of PQ+ = 81.06 . The code is at https://github.com/DTU-PAS/phenobench_challenge. Please download the weights in order to generate the predictions corresponding to the final submission.