Explainable ML Project 🐬

This project was created by Ilyesse and Gabriel as part of the Explainable Machine Learning module at the University of Applied Sciences Karlsruhe.

The dataset used in this project is the Animal Image Dataset from Kaggle, comprising 90 different animal species that needed to be classified.

The employed model is ResNet18, which was trained on the dataset using transfer learning techniques. Translation of animal names by deep-translator.

Usage 🦎

Predict: In the "Predict" tab, the model can be applied to high-resolution images to predict the species among the 90 different animals.

Explain: Under the "Explain" tab, the model can be applied to high-resolution images to obtain an explanation for the prediction. This explanation is generated using the Grad-CAM method.

Example Images: The "Example Images" section allows users to view sample images from the dataset. These images can be utilized as input by simply dragging and dropping them onto the interface. It is important to note that these example images were not part of the training data used for the model.