Some of the keypoints tests are done using the OCHuman dataset, but the parser for this dataset is not implemented in icevision (currently it's being defined inside conftest.py).
Instead we should do all tests using COCO keypoints, OCHuman should be moved to icedata.
Some of the keypoints tests are done using the OCHuman dataset, but the parser for this dataset is not implemented in
icevision
(currently it's being defined insideconftest.py
).Instead we should do all tests using COCO keypoints, OCHuman should be moved to
icedata
.