search

facebookresearch / CutLER

Code release for "Cut and Learn for Unsupervised Object Detection and Instance Segmentation" and "VideoCutLER: Surprisingly Simple Unsupervised Video Instance Segmentation"
Other
913 stars 90 forks source link

VOC annotations generated by maskcut #29

Closed zqwang21 closed 1 year ago

zqwang21 commented 1 year ago

Hey, this is my jsonfile for VOC annoatations generated by maskcut ,but the bbox information is different from the file you submitted in github. Therefore I visualize the results of both of us on JPEGImages/009829.jpg ，the yellow boxes are mine and the red are yours. image Do you have any idea why there are these differences? voc_annotations.zip

https://github.com/facebookresearch/CutLER/issues/28#issuecomment-1519094466

frank-xwang commented 1 year ago

Hi, the JSON file we provided contains ground truth annotations that have been converted to COCO-style format. The main purpose of providing this file is to evaluate the zero-shot unsupervised detection performance of our model. Also, the coco-style bounding box format should be [x_min, y_min, width, height], which is different from the voc format. Hope it helps!