Preliminary Evaluation of Machine Learning Approaches for PANTONE Color Classification with the SENSIPATCH wearable system

Precise color classification is crucial in industries such as art, design, manufacturing, textile, and biomedical applications. It ensures consistency, meets stringent standards, and aids in functional decision-making where accurate color distinction is essential. The SENSIPATCH wearable sensor system presents an end-to-end solution for spectral analysis with its multi-channel scattering spectrometer and SENSIPLUS microsensor, providing a seamless integration of sensor data acquisition, processing, and machine learning. This study investigates the efficacy of the SENSIPATCH sensor in accurately classifying PANTONE card colors. By leveraging spectral data, the study aims to enhance precision in color classification across various sectors and explore future biomedical applications, specifically melanoma detection. The primary objectives are to evaluate the SENSIPATCH sensor's accuracy in PANTONE color classification, develop and refine a Multi-Layer Perceptron (MLP) model to interpret the spectral data, and assess the potential of this technology for complex classification tasks in different fields. The methodology involves a controlled experimental setup, extensive data collection under varying lighting conditions, and the application of machine learning algorithms for color classification. Results demonstrate the superior performance of the MLP model in color classification compared to KNN, SVM, and Random Forest models.