Answers are grounded in this paper and cite their sources.
In this paper, we propose an effective cross-modal embedding fusing decoder (Cross Decoder) for cross-modal recipe retrieval tasks. We introduce our Cross Decoder into a recent GAN and transformer-based method to improve the representation capability of the recipe embeddings. By reconstructing images through GAN using embeddings learned by our Cross Decoder, we increase the reliability of embeddings, as well as achieving high-quality image generation. In addition, with dynamic margins adopted in the retrieval loss, the performance of the whole framework is further improved. The experimental results show that our method outperforms the state-of-the-art methods on the Recipe1M dataset.
No takes yet. Share an insight, caveat, or question.
Yang et al. (2024) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: