<![CDATA[In the era of Industry 4.0, the demand for manufacturing systems that are fast, precise, and efficient has become increasingly urgent. This drives the adoption of artificial intelligence (AI) technologies as a promising solution, including in the field of automatic bottle sorting. However, many industries still use manual bottle sorting systems, which often have significant drawbacks. This study presents an integrated artificial intelligence (AI)-based inspection model for automated bottle inspection in the context of smart manufacturing. The proposed approach integrates geometric feature extraction with region-of-interest (ROI)-based statistical image analysis to improve classification accuracy and robustness. Geometric features extracted from bottle contours are combined with optimized ROI selection to enhance feature relevance prior to classification using a Random Forest algorithm. The dataset consists of four bottle types: plastic, glass, cans, and cardboard, captured under controlled imaging conditions. Experimental results show that the proposed integrated method achieves classification accuracy ranging from 96% to 97.72%. The findings confirm that ROI optimization significantly influences statistical feature characteristics and improves overall model performance. This integrated framework is suitable for implementation in automated visual inspection systems supporting Industry 4.0 applications.]]>
Copyrights © 2026
