Multimodal large language models for food safety detection within deep learning frameworks: a review.

As the food industry continues to evolves and global trade expands, food safety challenges have become increasingly diverse, concealed, and frequent. Traditional methods such as cartographic analysis, mass spectrometry, and immunology assays offer high accuracy but suffer from long testing cycles, complex procedures, and low automation, limiting their effectiveness for high-throughput, intelligent supply chain monitoring. Recent advances in Multimodal Large Models (MLLMs) provide promising solutions by integrating multi-modal perception, knowledge enhancement, and self-supervised pre-training. Progress from deep learning to cross-modal intelligence and key multimodal fusion mechanisms are summarized, together with applications including quality assessment and upstream agriculture risks. Current challenges involving limited data resources, reliable intelligence generation, energy efficiency, and security are discussed, highlighting future directions for intelligent food safety detection.