pretrained model

[zeduchen]

Hello, thank you for open source your code. I want to run your code. Which pretrained model does your code use? Can you give me the download URL?

[HelloRicky123]

songdejia provides a pretrained model https://github.com/songdejia/Siamese-RPN-pytorch#download-pretrained-model-on-vid-with-690000-image-pairs

[zeduchen]

Thank you! But this pre-training model is not the same as your AlexNet.

[MrYANG23]

it seems like that the pretrained model of songdejia dont have the batchnormaliztion

[HelloRicky123]

During my training getting the 0.54 AUC, I use these code to load the former three layers of Li Bo's trained model, for they say these layers are frozen. " if config.pretrained_model: print("pre init checkpoint %s" % config.pretrained_model) checkpoint = torch.load(config.pretrained_model) model_dict = model.state_dict() keys_former3conv = ['featureExtract.0.weight', 'featureExtract.0.bias', 'featureExtract.4.weight', 'featureExtract.4.bias', 'featureExtract.8.weight', 'featureExtract.8.bias', ] checkpoint = {k: v for k, v in checkpoint.items() if k in keys_former3conv} model_dict.update(checkpoint) model.load_state_dict(model_dict) for k, v in model.named_parameters(): if k in keys_former3conv: v.requires_grad = False del checkpoint torch.cuda.empty_cache() print("pre inited checkpoint")"

[HelloRicky123]

And thanks to Li Bo's help, they give out the pretrained model on Imagenet, this is the address: https://drive.google.com/drive/folders/1HJOvl_irX3KFbtfj88_FVLtukMI1GTCR

[zeduchen]

Thank you!

[MrYANG23]

thank you !