In this project I was challenged to solve a unique set of 2000 captchas on a Raspberry-Pi. The solution involved designing a Deep-Neural-Network to train an image-classifying model. The trained model is then deployed on the Raspberry-Pi via Tensorflow Lite where the captchas reside.

Severel challenges are included in the captcha set including varying captcha lengths and illegal captcha symbols. The final model produced an 80% accuracy on the captcha set.