Research on faucet sensitivity detection method based on YOLOv5 and machine vision
To address the issues of large mechanical coupling errors, low detection accuracy, and insufficient automation in traditional faucet sensitivity detection methods, this paper proposes an intelligent detection method based on the YOLOv5 object detection algorithm and machine vision technology. This method first utilizes the YOLOv5 convolutional neural network for high-precision localization and identification of faucet handles. A specialized dataset encompassing various lighting and material conditions was constructed, and the Mosaic data augmentation strategy was employed to enhance the model's robustness under complex working conditions. Secondly, a non-contact high-precision measurement of the handle rotation angle was achieved based on the Hough Transform center detection algorithm. Finally, a complete detection system was designed and implemented. Experimental results show that the YOLOv5 model achieves a localization accuracy of 99.44%. The maximum measurement error for the handle rotation angle is 0.36°, with a measurement repeatability of 0.122° and a sensitivity detection repeatability of 0.146°. These results meet the requirements of the national standard GB18145-2014, which mandates a measurement error within ±0.5°. This study provides an efficient, accurate, and intelligent solution for faucet sensitivity detection.