I am trying to fine tune Deformable DETR on a custom dataset and I modified your DETR script by modifying the model and image processors from DETR to Deformable DETR. However, I am getting this error:
File "train_deformable_detr.py", line 367, in main
evaluator.update(predictions)
File "/miniconda/envs/myenv/lib/python3.8/site-packages/coco_eval/coco_eval.py", line 50, in update
img_ids, eval_imgs = evaluate(coco_eval)
File "/miniconda/envs/myenv/lib/python3.8/site-packages/coco_eval/coco_eval.py", line 126, in evaluate
self._prepare()
File "/miniconda/envs/myenv/lib/python3.8/site-packages/pycocotools/cocoeval.py", line 97, in _prepare
dts=self.cocoDt.loadAnns(self.cocoDt.getAnnIds(imgIds=p.imgIds, catIds=p.catIds))
File "/miniconda/envs/myenv/lib/python3.8/site-packages/pycocotools/coco.py", line 146, in getAnnIds
anns = self.dataset['annotations']
KeyError: 'annotations'
My code works perfectly fine for DETR though. I think there might be some API difference between Deformable DETR and DETR image processors. I appreciate any hints. Thank you!
Hi,
Thank you for the great collection of tutorials.
I am trying to fine tune Deformable DETR on a custom dataset and I modified your DETR script by modifying the model and image processors from DETR to Deformable DETR. However, I am getting this error:
My code works perfectly fine for DETR though. I think there might be some API difference between Deformable DETR and DETR image processors. I appreciate any hints. Thank you!