facebookresearch / ov-seg

This is the official PyTorch implementation of the paper Open-Vocabulary Semantic Segmentation with Mask-adapted CLIP.
Other
676 stars 61 forks source link

skip key renaming in _load_from_state_dict #1

Closed blumenstiel closed 1 year ago

blumenstiel commented 1 year ago

The keys containing "sem_seg_head.pixel_decoder" are changed to "sem_seg_head.pixel_decoder.pixel_decoder" which is not found in the provided model weights.