Load model from local path

[lanesket]

Is there any way to download the model (db_resnet50) from the directory (without Hub) if I already have the weights downloaded, this is needed to use the model on a computer without internet.

[lanesket]

predictor = detection_predictor("db_resnet50", pretrained=False)
weights = torch.load('../weights/db_resnet50-ac60cadc.pt', map_location=device)
predictor.model.load_state_dict(weights)

[Rilay9]

I'm trying to use the ocr predictor and it seems to try to download even if pretrained is set to false. Any thoughts?

[Rilay9]

Detection predictor also throws an error if offline.