Rapid Color Quality Evaluation of Needle-Shaped Green Tea Using Computer Vision System and Machine Learning Models

Area of Science:

• Agricultural Science
• Computer Vision
• Machine Learning

Background:

• Green tea quality, especially for needle-shaped varieties, relies heavily on color characteristics.
• Current quality assessment methods are subjective and rely on sensory analysis.
• There is a need for objective, precise, and efficient methods for evaluating tea color quality.

Purpose of the Study:

• To develop an objective methodology for assessing the color quality of needle-shaped green tea.
• To predict sensory evaluation results using computer vision and machine learning models based on color features.

Main Methods:

• Utilized computer vision technology to capture 885 images from 157 needle-shaped green tea samples.
• Employed three machine learning algorithms: Random Forest (RF), Support Vector Machine (SVM), and Decision Tree-based AdaBoost (DT-AdaBoost).
• Constructed color quality evaluation models using the extracted color features from the images.

Main Results:

• The Decision Tree-based AdaBoost (DT-AdaBoost) model demonstrated significant potential for tea quality evaluation.
• The DT-AdaBoost model achieved a correct discrimination rate (CDR) of 98.50% on 266 verification samples.
• The model also showed a relative percent deviation (RPD) of 14.827, indicating high accuracy and reliability.

Conclusions:

• The integration of computer vision and machine learning offers an effective approach for objective green tea color quality assessment.
• The DT-AdaBoost model provides a precise and efficient alternative to traditional subjective sensory analysis for needle-shaped green tea.
• This technology has the potential to revolutionize quality control in the green tea industry.