(Wo)man vs Machine Learning: Who Can Win the Image Classification Race?

The image classification demo challenges humans to ‘beat the machine’ in a race to identify a sequence of images. It uses a CIFAR-10 dataset, and a convolutional neural network (CNN) trained on a normal computer and deployed on an Arm Cortex-M7 CPU, to recognize 10 different image types. The Cortex-M7 takes input from the camera, pre-processes the image, and passes it to the CNN to identify the class of image. Running with CMSIS-NN – a library of open-source, low-level machine learning primitives targeting