-
Notifications
You must be signed in to change notification settings - Fork 7.9k
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
Caffe Master Branch Clip Layer prevents Openpose from loading model #799
Comments
We are using our own Caffe (that is basically a version that we know that works from master Caffe), so we are effectively using a specific tag. It works on my case with no problem, maybe re-downloading it and re-compiling it? |
Ah, you know, I've been using the main line Caffe, which worked until a couple of weeks ago. Your fork works fine. Apologies. |
Thanks for the feed-back. Best |
I was finally able to replicate the error and fix it. Our latest https://github.com/CMU-Perceptual-Computing-Lab/caffe Let me know if it still fails. |
So I got the same error: Clip Layer prevents Openpose from loading model. I am Openpose's version of caffee.
This is running the code from example1 |
Use the latest code with the default Caffe that OpenPose provides (i.e., remove your current repo, and re-clone it again so the default Caffe of OpenPose is updated). Thanks! |
Actually |
I am making a new issue for this, since OP closed #787.
Basically, Caffe Master recently added a clip layer that conflicts with a clip layer contained in the Body Pose 25 weights file. This causes a run-time failure when the model is loaded.
Reverting Caffe to BVLC/caffe@f019d0d fixed the issue for me.
I just changed from master to f019d0dfe86f49d1140961f8c7dec22130c83154 in the two relevant places:
openpose/CMakeLists.txt
Line 616 in f49e184
and
openpose/CMakeLists.txt
Line 651 in f49e184
It probably makes sense to switch to a specific tag or commit rather than master to avoid this type of issue in the future.
The text was updated successfully, but these errors were encountered: