Decomposing generation networks with structure prediction for recipe generation

作者：

Highlights：

• We divide the recipes into phases, and use a global structure prediction component to assign different subgenerators to generate recipe phases.

• We incorporate the attention mechanism to get the phase-aware features, which are the input for different subgenerators to produce better recipe phases.

• Our proposed framework outperforms previous state-of-the-art results.

摘要

•We divide the recipes into phases, and use a global structure prediction component to assign different subgenerators to generate recipe phases.•We incorporate the attention mechanism to get the phase-aware features, which are the input for different subgenerators to produce better recipe phases.•Our proposed framework outperforms previous state-of-the-art results.

论文关键词：Text generation,Vision-and-language

论文评审过程：Received 7 December 2020, Revised 2 August 2021, Accepted 6 February 2022, Available online 7 February 2022, Version of Record 20 February 2022.

论文官网地址：https://doi.org/10.1016/j.patcog.2022.108578