OPTIMIZING FEATURE EXTRACTION IN AI-BASED COCOON CLASSIFICATION: A HYBRID APPROACH FOR ENHANCED SILK QUALITY

Аннотация

Abstract. Silk production is a crucial industry that relies heavily on the classification of silk cocoons to ensure the highest quality output. Traditional classification methods are labor-intensive, inconsistent, and subject to human error. Artificial Intelligence (AI)-based systems, particularly those using deep learning, have significantly improved classification accuracy. However, optimizing feature extraction techniques remains a challenge. This paper explores a hybrid approach that combines deep learning with traditional machine learning feature extraction methods to enhance classification accuracy and silk quality assessment [1][2].

This study examines various feature extraction techniques, including texture, shape, and color analysis, alongside CNN-based automatic feature selection. By integrating handcrafted features with AI-driven feature learning, we propose a robust classification system that improves efficiency and accuracy. The proposed approach is validated using real-world datasets, and its implications for large-scale silk production are discussed [3].