The first video shows my first experiment of live recognition using a Pi3, USB webcam, Movidius\Caffe and OpenCV on a pre-trained neural net called SqueezeNet.
Now that we have a successful benchmark for Movidius, we will turn our efforts to benchmarking without it, and using YOLO\Darknet instead of Caffe\Squeezenet.
After that, we will train\compile a custom neural net on traffic-related images which should achieve a faster recognition speed on either platforms. The goal is < 40 ms on a small custom set of objects.
Discussions
Become a Hackaday.io Member
Create an account to leave a comment. Already have an account? Log In.