Testing of NN accelerated inference of age/gender recognition model on reTerminal, based on CM 4,~16-20 FPS. The results are promising, will continue tweaking the model architecture and refine the data.
Conversation
Thanks! It's a bit challenging on latest Raspberry Pi OS Buster - since pre-built binaries are only available for Bullseye. I haven't posted an issue on ARM NN Github, since it's not code related. Is there anybody I can talk to about binaries/setting up process?
1
Show replies
Could you share any documentation you've made for this? This is really cool Btw. Inference on Rpi4 in Real-time for Age Gender Recognition NN is super impressive.
2
I'm still tweaking the model, want to achieve optimal balance between inference speed and accuracy. Once I'm done I'll upload the training code to github.com/Seeed-Studio/S the inference code is already there
1
1
1
Show replies