- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
I had tested OpenVINO Action Recognition using Kinetics 400, but for actions such as violence, Moments In Time from MIT-IBM WATSON seems more accurate, I have tested using videos from YouTube.
Moments in Time is based on Inception V2 from the following GitHub page :
https://github.com/metalbubble/TRN-pytorch, Both models are based on Python.
Can you please let me know the process to convert the Pytorch model to OpenVINO, as nothing can beat OpenVINO on Intel processors.
Or
Can I just replace the Kinetics 400 with Moments In Time Dataset and test?
Please share me your thoughts.
Thanks
Guru
Link Copied
1 Reply
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi, Guruvishnuvardan,
For Pytorch model, you need to convert it to ONNX then to OpenVINO IR. As for using the other dataset with the original Action Recognition model, you have to re-train the model with the dataset, you can follow the guidance of OpenVINO training extension attached below to train the model with the your own dataset.

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