Machine Learning Projects Page

Check out my github @SmallChungus1. For source codes! Check out my dropbox For trained models ready to use. Thanks for checking them out!

Forest Fire Image Binary Classifier

Built a binary image classification CNN implementing the VGG-16 architecture using Keras. Used to predict if the given image depicts forest fire or not. Achieved a 90% accuracy on test batches. Packaged into a web app with Flask to allow for image classifiation with image urls on the web. Click into this to learn more!

Playing Cards Classifier

Designed and trained a lightweight 6-layer CNN architecture to classify a deck of playing cards. Achieved 91% detection accuracy with 30 epochs.

Car Image Segmentation

Trained a U-Net model for car image segmentation task. Achieved dice score of 0.98