EvalDeepPose.py - Assign requires shapes of both tensors to match. lhs shape= [20] rhs shape= [48]

[pbanavara]

Guys - This is not a bug. Just a question that I want clarification on. I have trained the model and want to run it against my own data. My own data contains 147 images. I have resized the images to 100w * 100h to match the training data. When I run EvalDeepPose.py I get this error

Assign requires shapes of both tensors to match. lhs shape= [20] rhs shape= [48]

I am not sure how this size is computed. I printed out all the variables and here is that output.

ensor("Placeholder:0", shape=(128, 100, 100, 3), dtype=float32) (11, 11, 3, 20) conv1/weights:0 (20,) conv1/biases:0 (5, 5, 20, 35) conv2/weights:0 (35,) conv2/biases:0 (3, 3, 35, 50) conv4/weights:0 (50,) conv4/biases:0 (3, 3, 50, 75) conv5/weights:0 (75,) conv5/biases:0 (300, 1024) local1/weights:0 (1024,) local1/biases:0 (1024, 1024) local2/weights:0 (1024,) local2/biases:0 (1024, 0) softmax_linear/weights:0 (0,) softmax_linear/biases:0

So I see the LHS but I can't understand where the RHS shape [48] is coming from. I have not changed input_size in LSPGlobals.py. I have just changed the input directory parameter to a new directory containing all new images. Any help in resolving this will be greatly appreciated. I'll post this on SO as well.

[samitok]

Hello, sorry to inform you, but this code is not working well. I will make a big update in the following weeks.

[pbanavara]

No worries. I'm more than happy to collaborate. If you have any specific issues that you want resolved - list them out and I'll submit pull requests accordingly.

[samitok]

Thank you for your willingness. I may list some isssues after I make the important changes.