This is an image classifier I made in 2020 to get used to different aspects of tensorflow and keras' machine learning library. There's countless youtube tutorials about quick, accurate classifiers with surefire models to train, but I wanted to see how far I could get with tensorflow's datasdets. I chose the 'rock paper scissors' dataset in which we try to train a calssifier to detect the three hand gestures.

After achieving 82% accuracy after a convolutional neural network approach using perameter hypertuning, I decided to keep my failed attempts as seperate cells in a google colab, so that I can retrace my steps whenever I have to work in other fields to come back to it. Tensorflow and Keras are great for their high-level methods making the frameworks very accessible. Because it's so accessible, I actually have a lot of fun with image classifiers with mixxed success. Every once in a while I come back to this and play with the numbers or add a method to see if I can get the accuracy up in the hypertuning parameter test phase. It's actually really fun so if you have an idea on how to make it better let me know!