BURST-benchmark
BURST-benchmark copied to clipboard
file naming in annotations.zip
annotations.zip has the following files.
- train:
- train.json
- val:
- all_classes.json
- common_classes.json
- uncommon_classes.json
- first_frame_annotations.json
- test:
- all_classes.json
- common_classes.json
- uncommon_classes.json
- first_frame_annotations.json
- info:
- categories.json
- class_split.json
My question is: in train, does train.json is equivalent to all_classes.json or I downloaded old or wrong annotation?
Hi, yes for the training set all class annotations are in a single file. The separate files for val/test are sort of a legacy artifact from earlier when we needed to point the eval script to different files depending on which class split we wanted to evaluate. This is however no longer the case.