Colonoscopy Image Analysis with UNet

This repository presents a comprehensive analysis of colonoscopy images using the UNet architecture for both classification and segmentation tasks.

Tasks

1. Classification:

   • Trained a Resnet model for classifying colonoscopy images into different categories.
   • Achieved high accuracy in distinguishing between various classes, contributing to improved diagnostic capabilities.

2. Segmentation:

   • Implemented UNet for semantic segmentation of colon regions in the images.
   • Generated precise masks highlighting the colon structures, aiding in detailed analysis and pathology detection.

Results

Model Trained From Scratch

Model Trained using Pretrained weights