Object Segmentation

Designed a custom Convolutional Neural Network to perform semantic segmentation on a custom dataset to segment roads and vehicles. Custom CNN has 5 layers and skip connections for encoder and decoder models, and achieved 93% accuracy.