-
Notifications
You must be signed in to change notification settings - Fork 109
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
ONNX output is giving incorrect DOF values #77
Comments
have you soloved it ?I encountered this issue too. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Hi Vítor, I tried to convert this model to onnx. When I tried running inference using this model, the bounding box and accuracy was very similar and within acceptable range but when I tried to run the align function the output was very distorted. And I found out that the DOF output was giving weird output. Here's an example of the
onnx model dof
[0.371725 -0.275935 1.866960 0.021321 -0.280082 0.334366]
main model dof
[0.083832 -0.089648 0.311040 0.715069 -0.212262 10.305621]
I'm not sure what could cause this problem. I modified the run_face_align.py file to convert the instance of the model to onnx and also infer in the same file. I'm pasting the contents of the file here for additional context. You can call the align function by passing different parameters to run different functions. I tried to modify the aligning with something else in the "aligning_faces_onnx_new" section. That one was purely by luck though but I'm interested finding the problem and fixing the onnx output.
Also here's my conda env info in short
torch: 1.7.1, python: 3.9.16, onnx: 1.14.0,
Let me know if you have any further questions.
Thank you
The text was updated successfully, but these errors were encountered: