Classification of Cardamom Samples Based on Concentration of Important Biochemical Constituents Using InceptionV3-Driven Computer Vision System

Cardamom, a versatile spice celebrated for its unique aroma and taste, boasts a rich and captivating history. Different grades of cardamoms are commercially available, for example green, black, and white cardamoms. Again within each of these, there are variety of cardamoms in terms of their biochemical constituents and quality which govern their pricing in market as well. Conventionally, different spectral and instrumental techniques are used for grading of cardamoms. This paper proposes a computer vision approach for the same whilst using the potential of InceptionV3 convolutional neural network. Four different cardamom grades have been considered for experiments, and results have shown 97% accuracy in identification of cardamom grades.