- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
iam trying to compile the MTCNN network from davidsandberg (https://github.com/davidsandberg/facenet/blob/master/src/align/detect_face.py). While compiling iam getting this error: [Error 5] Toolkit Error: Stage Details Not Supported: Neg
What operation is ment?
Link Copied
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
@daku Thank you for reporting this issue. Unfortunately, it seems like we don't quite have support for that network yet or more specifically, the tf.neg() or tf.negative() operation. We will make a note of this operation and network and will consider support for the tf.neg/negative operation for a future release.
We do have a facenet example located in the ncappzoo github repository if you are interested. It is based on a model from David Sanberg as well. Link to the ncappzoo (https://github.com/movidius/ncappzoo)
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
@daku Also can you provide a link to your prototxt and caffe model for testing and debugging?
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I cannot find any tf.neg or tf.negative function in the detect_face.py.
Iam trying to compile the implementation of MTCNN Network in tensorflow for detecting faces in pictures. It can be found here: https://github.com/davidsandberg/facenet/tree/master/src/align. It is a different neural net in camprison to the facenet.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
@daku Thanks for providing a link to your network. I could not find any tf.neg() operations in that github as well. I wonder if there is some composite function call that calls tf.neg(). Will check the network, but most likely this is something that we will have to add in a future release.

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page