KelvinSun

执行bash test_mask_refine.sh config_vot.json SiamMask_VOT.pth VOT2016 0的问题

5

VOT2016下载并解压在data目录下，预训练模型也下载了。但是执行这个语句还存在如下问题 usage: test.py [-h] [--arch {Custom}] --config CONFIG --resume PATH [--mask] [--refine] [--dataset {}] [-l LOG] [-v] [--save_mask] [--gt] [--video VIDEO] [--cpu] [--debug] test.py: error: argument --dataset: invalid choice: 'VOT2016'...

Could the training batchsize turned up

1

When I turn up the value of train batch, it will appear this **masks = torch.stack([sample[1] for sample in batch]) RuntimeError: stack expects each tensor to be equal size, but...