Must provide as many biases as the channel dimension of the input tensor
murugeshmanthiramoorthi opened this issue · 1 comments
I have trained the model on my own data with same size 192 x 192. The whole training was completed without any errors. I saved my best fit model. But when I try to load the model for predicting, I get the following error.
InvalidArgumentError: Must provide as many biases as the channel dimension of the input tensor: [3] vs. 9 in [1,9,192,192]
[[node StackedDenseNet/StackedDenseNet/image_net_preprocess_1/BiasAdd (defined at /usr/local/lib/python3.6/dist-packages/deepposekit/models/layers/deeplabcut.py:86) ]] [Op:__inference_predict_function_341865]
Errors may have originated from an input operation.
Input Source operations connected to node StackedDenseNet/StackedDenseNet/image_net_preprocess_1/BiasAdd:
StackedDenseNet/StackedDenseNet/image_net_preprocess_1/truediv (defined at /usr/local/lib/python3.6/dist-packages/deepposekit/models/layers/deeplabcut.py:62)
StackedDenseNet/StackedDenseNet/image_net_preprocess_1/Const (defined at /usr/local/lib/python3.6/dist-packages/deepposekit/models/layers/deeplabcut.py:78)
Function call stack:
predict_function
I think this would solve that issue.
#29