search

DerrickXuNu / OpenCOOD

[ICRA 2022] An opensource framework for cooperative detection. Official implementation for OPV2V.
https://mobility-lab.seas.ucla.edu/opv2v/
Other
674 stars 100 forks source link

ego cavs are not in the ground truth, will it affect the acc? #141

Closed drowning-in-codes closed 2 months ago

drowning-in-codes commented 4 months ago

Hi, great thanks to your code base! I'm a bit confused about some settings in the code. In the opv2v data yaml files, vehicles are objects that collaborative perception will detect(i.e. gtbox), and I find that there are no ego and support cavs. But the predicted result will usually output the ego and support cavs bbox, so there is always a mismatch. Will it affect the iou or acc? Thanks again for your great work and looking forward to your reply.

xiaojianyuyong commented 2 months ago

Same problem, have you solved this problem?

DerrickXuNu commented 2 months ago

Yes, it will actually influence the AP a bit. But since all the methods use the same setting, i guess it is fair? You are welcome to remove those or add gt for it.