This paper presents Discriminative Part Network (DP-Net), a deep architecture with strong interpretation capabilities, which exploits a pretrained Convolutional Neural Network (CNN) combined with a part-based recognition module. This system learns and detects parts in the images that are discriminative among categories, without the need for fine-tuning the CNN, making it more scalable than other part-based models. While part-based approaches naturally offer interpretable representations, we propose explanations at image and category levels and introduce specific constraints on the part learning process to make them more discrimative.
@article{arxiv.2404.15037,
title = {DP-Net: Learning Discriminative Parts for image recognition},
author = {Ronan Sicre and Hanwei Zhang and Julien Dejasmin and Chiheb Daaloul and Stéphane Ayache and Thierry Artières},
journal= {arXiv preprint arXiv:2404.15037},
year = {2024}
}