how to run the models against a video?

[frandoLin]

I am bit confused about the final result by running the main.py when inferencing. Are they just a couple of feature embeddings for the input images?

[VlSomers]

Hi @frandoLin, indeed, a reid module produces a feature embeddings for each input image (multi part-based embeddings in the case of BPBreID). You can then use these embeddings to perform retrieval in a gallery or tracking in a video. You can have a look at the torchreid/models/bpbreid.py class to see the output dictionary that contains the embeddings. LEt me know if this answers you question.