- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Does openvino support asymmetric convolution? I need to split my 5x5 kernel into 5x1 and 1x5
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
hello. I have already found that the openvino supports the asymmetric convolution. and it was also used in inception v3
Thank you
Link Copied
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Shirleyliu,
Thanks for reaching out to us.
Currently, I do not notice any model from Open Model Zoo are using this asymmetric convolution.
Do you have any other information to provide for further understanding about this asymmetric convolution?
Regards,
Peh
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
hello. I have already found that the openvino supports the asymmetric convolution. and it was also used in inception v3
Thank you
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Shirleyliu,
Thanks for sharing back your findings to us.
Just to add comment on your sharing, googlenet-v3-pytorch model is the model that use asymmetric convolutions. This PyTorch implementation of the architecture is described in the paper “Rethinking the Inception Architecture for Computer Vision”
This thread will no longer be monitored since you have answered your question. If you need any additional information from Intel, please submit a new question.
Regards,
Peh
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page