Using neural network to ensure the packing quality of pasteurized milk

The article considers the application of the neural network model to check the quality of packaging of drinking milk in the technological production process. The neural network allows you to remove the routine load from the verification staff. Method of detection of possible defective areas and identification of signs of damage of packing areas is used. Experiments have shown that when using the proposed solution in production - the accuracy indicator can exceed 80%. Detecting scrap on a conveyor belt in order to detect objects of a certain class is one of the most complex branches of computer vision. Methods of existing standard detection models were analyzed, with the development of which the performance of deep learning networks was significantly improved. An overview of systematic object detection methods covering single-stage and two-stage detectors is provided. Use of these methods of object detection for creation of efficient neural network technology is considered.