This is an Image Classifier that i created while doing the fast.ai course, it makes use of the fast.ai libraries to implement a RESNET 50 model which was trained on images downloaded using Bing Images. the classifier can differentiate between single sculling boats and quads (which have four scullers in each boat) with 90% accuracy on a self labelled validation set of 100 images.
There is a simple web app built using ipython widgets and published on binder here: https://mybinder.org/v2/gh/timjones1/rowing-classifier/master
A short article on my experience implementing this model and web app is here: https://timjones1.github.io/post/rowing-classifier/