Fusion Learning using Semantics and Graph Convolutional Network for Visual Food Recognition

• Focus on food recognition in the domain of fusion learning, which combines few-shot and many-shot learning.
• This project explores using semantic information, such as class label textual embedding or hierarchical embedding to improve the image classification performance.
• Inter-class correlations in terms of image feature and text representation are explored using a designed graph convolutional network.
• The proposed method achieves state-of-the-art performance on major food benchmark datasets.

• Published in WACV2021.

  Overview of the Proposed Fusion Learning Framework for Food Recognition

For more information, you may contact our professor Yap Kim Hui.

Back to Research Areas