Using our own dataset for testing

[xertxiao]

Is there a way to prepare our own dataset and test the models on it? Thanks

[KaiyangZhou]

Yes, you could follow .py files in dataset_manager/ and write your own dataset classes.

[xertxiao]

Thanks!

[xertxiao]

Hi, I was trying to use a pre-trained market1501 model on my own dataset but failed. The error messages says that this pre-trained model expect a "torch.Size([751])" (I suspect this 751 is the total ids of people in market) but my dataset has "torch.Size([16])", where 16 is total ids of my dataset. Is there anything I can do to resolve this? Thanks again.

[xertxiao]

@KaiyangZhou thanks!

[KaiyangZhou]

use --load-weights.

[xertxiao]

Thanks for helping!