The relevance of accurately predicting fat content on the surface of meat products is driven by the need to ensure effective quality control in the food industry. This paper presents an efficient method for detecting and quantitatively assessing fat content on meat surfaces based on the use of color segmentation in the HSV color space. The method exploits differences in the color characteristics of fat and muscle tissues to effectively segment images of meat samples, calculate the percentage of fat, and analyze its spatial distribution. The simplicity and robustness of the algorithm make it a promising solution for real-time automated quality control systems, offering ease of use and high computational efficiency.

Keywords: computer vision, color segmentation, HSV color space, image processing, OpenCV, Flask, quality control automation

This paper addresses the problem of improving the accuracy and objectivity of meat quality assessment. An approach to automated product evaluation is presented, based on the integration of a computer vision system (CVS) and deep learning methods. A convolutional neural network VGG-16, pre-trained on the large-scale ImageNet dataset, was employed to effectively implement transfer learning. The processes of data preprocessing and the architecture of the applied neural network are described. The results of training and validation demonstrate high classification accuracy of meat samples by the criterion “fresh/spoiled.” The potential of the proposed approach for automating quality control in the meat industry and reducing the influence of subjective factors is emphasized.

Keywords: computer vision, deep learning, convolutional neural network, transfer learning, meat quality assessment, process automation, image classification, automated quality control

The paper presents an intelligent control system for the indirect assessment of fruit damage volume based on the use of a computer vision system and a convolutional neural network (CNN). An algorithm has been developed that analyzes the surface defect area to predict the volume of damaged pulp. The proposed approach includes stages of image acquisition, preprocessing, defect segmentation using a CNN, regression-based damage volume estimation, and decision-making based on fuzzy logic. A mathematical model is described that links the defect area to the damage volume, taking into account the internal spread of rot within the fruit. The presented system enables prompt and objective quality control of fruits, contributing to the optimization of sorting, storage, and processing operations in the food industry and the agro-industrial sector.

Keywords: intelligent control system, technical vision, sorting automation, convolutional neural network, image segmentation, fruit damage assessment, fuzzy logic