Machine learning, in combination with optical sensing, extracts key features from high-dimensional data for non-destructive food quality assessment. This approach overcomes the limitations of traditional destructive and labor-intensive methods, facilitating real-time decision-making for food quality profiling and robotic handling. This mini-review highlights various optical techniques integrated with machine learning for assessing food quality, including chemical profiling methods such as near-infrared, Raman, and hyperspectral imaging spectroscopy, as well as visual analysis such as RGB imaging. In addition, the review presents the application of robotics and computer vision techniques to assess food quality and then drives the automation of food harvesting, grading, and processing. Lastly, the review discusses current challenges and opportunities for future research.

Machine learning, in combination with optical sensing, extracts key features from high-dimensional data for non-destructive food quality assessment. This approach overcomes the limitations of traditional destructive and labor-intensive methods, facilitating real-time decision-making for food quality profiling and robotic handling. This mini-review highlights various optical techniques integrated with machine learning for assessing food quality, including chemical profiling methods such as near-infrared, Raman, and hyperspectral imaging spectroscopy, as well as visual analysis such as RGB imaging. In addition, the review presents the application of robotics and computer vision techniques to assess food quality and then drives the automation of food harvesting, grading, and processing. Lastly, the review discusses current challenges and opportunities for future research.