Instruments and Systems: Monitoring, Control, and Diagnostics Annotation << Back Automated Control of Oil Pollution on Rivers in the Oil Refining Process using Machine Vision and Neural Network Segmentation Methods K.D. SKOBELEV, V.G. BLAGOVESHCHENSKY An automated system for monitoring oil pollution in rivers has been developed using computer vision methods and neural network segmentation. The YOLO neural network was used as the core model, trained on a hybrid dataset comprising both real and synthetically generated images of oil contamination. The system enables accurate detection of oil fi lms on the water surface under challenging lighting and background conditions. High detection accuracy was achieved, ensuring near real-time operation. An integration architecture is proposed to embed the model into automated process control systems (APCS) of oil refi neries for proactive environmental monitoring and prevention of pollution spread. The solution is intended for practical use in high-risk areas where timely detection of leaks and mitigation of environmental consequences are required. Keywords: automated control, oil pollution, machine vision, neural network, YOLO, segmentation, monitoring. Pp. 31-37.