I just thought I’d share my success training 2,500 images on bvlc_googlenet using AWS and deployment on Jetson TX2. The test is made on a fresh video not used for training:
[url]Wasp Video Test Jetson TX2 - YouTube
More info is here:
[url]Comparison Between Mobilenet SSD and bvlc Googlenet | Details | Hackaday.io
If anybody from Nvidia happens to read this, please feel free to send me a Xavier (or two) !!