Tried to load custom Models on Tx2 but imagenet-console did not use extended command line parameters.
Still loaded googlenet.caffemodel and got the same classified result.
Here are some steps I did.
Is there I missed?
- retained a ILSVRC12-subset on DIGITS by followed Re-training the Network of two days to a demo.
- Tested a robin image on DIGITS and got 100% bird.
- Downloading Model snapshot to Jetson.
- run the imagenet-console with extended parameter.
- result was a 99% Robin still.
Thank you for any suggestions.