Decomposing Generation Networks with Structure Prediction for Recipe Generation

TL;DR

This paper introduces a novel framework for recipe generation that decomposes cooking instructions into phases using structure prediction, leading to more accurate and complete recipe outputs from food images and ingredients.

Contribution

The paper proposes a new method that decomposes recipe generation into phases with structure prediction, improving over previous models in capturing recipe structure and details.

Findings

Improved performance on Recipe1M dataset

Effective decomposition of instructions into phases

Enhanced structure understanding in recipe generation

Abstract

Recipe generation from food images and ingredients is a challenging task, which requires the interpretation of the information from another modality. Different from the image captioning task, where the captions usually have one sentence, cooking instructions contain multiple sentences and have obvious structures. To help the model capture the recipe structure and avoid missing some cooking details, we propose a novel framework: Decomposing Generation Networks (DGN) with structure prediction, to get more structured and complete recipe generation outputs. Specifically, we split each cooking instruction into several phases, and assign different sub-generators to each phase. Our approach includes two novel ideas: (i) learning the recipe structures with the global structure prediction component and (ii) producing recipe phases in the sub-generator output component based on the predicted structure. Extensive experiments on the challenging large-scale Recipe1M dataset validate the effectiveness of our proposed model, which improves the performance over the state-of-the-art results.