Accurate food nutrition estimation from single images is challenging due to the loss of 3D information. While depth-based methods provide reliable geometry, they remain inaccessible on most smartphones because of depth-sensor requirements. To overcome this challenge, we propose PortionNet, a novel cross-modal knowledge distillation framework that learns geometric features from point clouds during training while requiring only RGB images at inference. Our approach employs a dual-mode training strategy where a lightweight adapter network mimics point cloud representations, enabling pseudo-3D reasoning without any specialized hardware requirements. PortionNet achieves state-of-the-art performance on MetaFood3D, outperforming all previous methods in both volume and energy estimation. Cross-dataset evaluation on SimpleFood45 further demonstrates strong generalization in energy estimation.
@article{arxiv.2512.22304,
title = {PortionNet: Distilling 3D Geometric Knowledge for Food Nutrition Estimation},
author = {Darrin Bright and Rakshith Raj and Kanchan Keisham},
journal= {arXiv preprint arXiv:2512.22304},
year = {2025}
}
Comments
Accepted at the 11th Annual Conference on Vision and Intelligent Systems (CVIS 2025)