This repository contains the software for the EndoMapper dataset validation described in the section "Anatomical region validation" of the paper:
Azagra P. et al. Endomapper dataset of complete calibrated endoscopy procedures. Scientific Data. 2023. Accepted for publication
- Download the EfficientNet model from the Github and add it into the efficient folder.
- Download the checkpoint for the EfficientNet ckpt and unzip it in the efficientnetv2-m folder.
- Download the Anatomical Regions files from the Endomapper dataset and the video into the Data Folder.
- Extract the data with the Extract.sh script and create the dataset structure using the Python script CreateDataset.py
- To train a model: python train_t.py trained_efficient efficient
- To test a Model: python Test.py trained_efficient efficient