14: The Hyper Kvasir Dataset Landmark Classification (Upper GI tract)

This model was trained to identify anatomical landmarks in the upper GI tract during an endoscopy. Training was done on the Hyper-Kvasir dataset, which is the largest publicly released gastrointestinal tract image dataset. In total, the dataset contains 110,079 images and 373 videos where it captures anatomical landmarks and pathological and normal findings. The results is more than 1,1 million images and video frames all together. The paper describing the data can be accessed here. Here you will find the files used to prepare the dataset, create the baseline experiments, and the official k-fold splits of the dataset. Available classes are

• ileum
• retroflex-rectum

The model was trained using this script; a summary of its performance can be found here.