The goal of this project was to create a classifier that is able to distinguish between different types of bears, namely grizzly, black, and teddy bears. To achieve this, I used Tensorflow 2.0 and MobileNetV2 with imagenet weights, and a data sample of 269 images of various bears pulled from google search. The result was a model with 100% validation and testing accuracy.
Training and validation accuracy and loss graphs shown below.
See the source on github!