<![CDATA[In the canned sardine production industry, sealing issues often arise due to various factors, such as the quantity of fish in the can or improper calibration of the sealing machine. These sealing defects can result in poorly sealed cans that may explode and contaminate an entire production batch, leading to significant financial losses and damage to the company's reputation. This study proposes an advanced and reliable method for classifying fish can images to detect potential defects, such as sealing issues, which are critical to maintaining quality standards in the canning industry. Our classification method utilizes the Local Binary Patterns (LBP) algorithm for feature extraction across the entire dataset of images. The extracted features are then processed using a Perceptron classifier to identify poorly sealed cans. This approach achieved a precision score of 0.85, demonstrating its effectiveness. Additionally, our analysis revealed that LBP significantly contributes to improving classification accuracy. By automating and enhancing the quality assurance process, this method provides the canning industry with a robust tool for ensuring high product standards, minimizing errors, and increasing efficiency in production lines.]]>
Copyrights © 2025
