Objective To address the low efficiency and strong subjectivity of traditional manual tomato grading, this study developed an online tomato internal and external quality detection and grading system based on the Internet of Things (IoT) and machine learning technologies, enabling real-time, non-destructive detection of both internal and external quality attributes.Methods By integrating machine vision and near-infrared spectroscopy, and leveraging IoT and machine learning algorithms, a comprehensive system for online, non-destructive tomato detection and grading was designed and implemented. Real-time acquisition of external images and internal spectral information of tomatoes was performed. External defects, shape index, and diameter were detected using deep learning models, while soluble solids content and firmness were predicted using near-infrared spectroscopy. Ultimately, this enabled online detection and grading of tomato quality.Results The system demonstrated excellent performance: the accuracy of external quality detection reached 94.9%, internal quality prediction accuracy was 87.3%, and the integrated grading accuracy improved to 88.5%. The system achieved a processing efficiency of 19 tomatoes per minute.Conclusion By synergistically optimizing machine vision and near-infrared spectroscopy, the system overcomes the limitations of traditional single-attribute detection approaches, significantly improving the accuracy and efficiency of internal and external quality grading of tomatoes.