Problemas encontrados al ejecutar SiamMask

Problemas encontrados:

THCudaCheck FAIL file=/pytorch/aten/src/THC/THCGeneral.cpp line=663 error=11 : invalid argument
Traceback (most recent call last):
  File "../../tools/demo.py", line 53, in <module>
    state = siamese_init(im, target_pos, target_sz, siammask, cfg['hp'], device=device)  # init tracker
  File "/home/pwh/SIAMMASK/SiamMask/tools/test.py", line 155, in siamese_init
    net.template(z.to(device))
  File "/home/pwh/SIAMMASK/SiamMask/experiments/siammask_sharp/custom.py", line 174, in template
    self.zf = self.features(template)
  File "/home/pwh/.conda/envs/siammask/lib/python3.6/site-packages/torch/nn/modules/module.py", line 477, in __call__
    result = self.forward(*input, **kwargs)
  File "/home/pwh/SIAMMASK/SiamMask/experiments/siammask_sharp/custom.py", line 59, in forward
    output = self.features(x)
  File "/home/pwh/.conda/envs/siammask/lib/python3.6/site-packages/torch/nn/modules/module.py", line 477, in __call__
    result = self.forward(*input, **kwargs)
  File "/home/pwh/SIAMMASK/SiamMask/experiments/siammask_sharp/resnet.py", line 218, in forward
    x = self.conv1(x)
  File "/home/pwh/.conda/envs/siammask/lib/python3.6/site-packages/torch/nn/modules/module.py", line 477, in __call__
    result = self.forward(*input, **kwargs)
  File "/home/pwh/.conda/envs/siammask/lib/python3.6/site-packages/torch/nn/modules/conv.py", line 301, in forward
    self.padding, self.dilation, self.groups)
RuntimeError: cuda runtime error (11) : invalid argument at /pytorch/aten/src/THC/THCGeneral.cpp:663

Solución:

La razón principal de este problema es que la versión de la antorcha no se corresponde pip install -r requirment.txtcon ella, la versión de la antorcha se descargó originalmente 0.4.0
. Entonces se puede pip install -U torchactualizar (pero es mejor especificar la versión correspondiente a la versión cuda de su propio sistema). Cabe señalar aquí que 2021年torch版本已经更新到1.8，si ingresa directamente el comando anterior, la antorcha se actualizará a la versión 1.8, por lo que encontré el siguiente error:

Traceback (most recent call last):
  File "../../tools/demo.py", line 53, in <module>
    state = siamese_init(im, target_pos, target_sz, siammask, cfg['hp'], device=device)  # init tracker
  File "/home/pwh/SIAMMASK/SiamMask/tools/test.py", line 155, in siamese_init
    net.template(z.to(device))
  File "/home/pwh/SIAMMASK/SiamMask/experiments/siammask_sharp/custom.py", line 174, in template
    self.zf = self.features(template)
  File "/home/pwh/.conda/envs/siammask/lib/python3.6/site-packages/torch/nn/modules/module.py", line 889, in _call_impl
    result = self.forward(*input, **kwargs)
  File "/home/pwh/SIAMMASK/SiamMask/experiments/siammask_sharp/custom.py", line 59, in forward
    output = self.features(x)
  File "/home/pwh/.conda/envs/siammask/lib/python3.6/site-packages/torch/nn/modules/module.py", line 889, in _call_impl
    result = self.forward(*input, **kwargs)
  File "/home/pwh/SIAMMASK/SiamMask/experiments/siammask_sharp/resnet.py", line 218, in forward
    x = self.conv1(x)
  File "/home/pwh/.conda/envs/siammask/lib/python3.6/site-packages/torch/nn/modules/module.py", line 889, in _call_impl
    result = self.forward(*input, **kwargs)
  File "/home/pwh/.conda/envs/siammask/lib/python3.6/site-packages/torch/nn/modules/conv.py", line 399, in forward
    return self._conv_forward(input, self.weight, self.bias)
  File "/home/pwh/.conda/envs/siammask/lib/python3.6/site-packages/torch/nn/modules/conv.py", line 396, in _conv_forward
    self.padding, self.dilation, self.groups)
RuntimeError: cuDNN error: CUDNN_STATUS_NOT_INITIALIZED

Esto es causado cuda、cudnn、pytorchpor la versión incompatible, así que desinstalé la versión 1.8 de antorcha y elegí la versión 1.7 de antorcha. El comando es el siguiente :,
pip install -i https://pypi.douban.com/simple/ torch==1.7el problema se resuelve con éxito después de la actualización.