This repository contains the entire dataset of the endoscopic images. The dataset has the following components
a. Data b. Pre-processing c. Training
a. Data - The data folder has the following components
- Annotations - The annotation folder contains a collection of JSON files for both polp and ulcer. Every image has a separate annotation.json file. The format of the annotation is following the stabdard Coco format.
- Crops - This folder has all the crops and its mainly created to quickly recognition tasks
- Data_set - This folder has the entire curated dataset with train and test split for easy plug and play training
- raw_images - This folder contains raw images for the respective labels.
b. Pre-processing - This folder contains python files which help in pre-processing raw images and integrating annotation files with raw images.
c. Training - This folder contains training python files for image recognition tasks.