Deep learning-based phenotypic analysis and intelligent environmental control in edible mushrooms: Advances, challenges, and prospects

Edible mushrooms are gaining global recognition for their nutritional value, ecological benefits, and economic importance. However, their cultivation practices remain largely empirical, posing challenges for standardized breeding, quality control, and environmental regulation. Recent advances in deep learning offer new opportunities to address these issues by enabling high-throughput phenotypic analysis and intelligent environmental control. This review systematically summarizes the current progress in applying deep learning techniques across multiple dimensions of mushroom cultivation, including two-dimensional image-based morphology analysis, three-dimensional reconstruction, spectral feature fusion, and time-series growth modeling. We further discuss the development of environment–phenotype coupling frameworks and closed-loop control systems empowered by multimodal sensing, edge computing, and digital twin technologies. While these innovations mark a significant shift toward data-driven cultivation, we also identify persistent challenges such as data scarcity, model generalization, and deployment under real-world constraints. Finally, we highlight future directions including cross-modal self-supervised learning, federated learning, and variety-specific adaptive modeling. This review provides a comprehensive roadmap for advancing intelligent mushroom production through deep learning integration.