Metro air valve state dataset for machine learning monitoring and diagnosis

The Images were collected and shot in Jinan Metro, China from May 2022 to August 2022, and then obtained by video frame processing. After various image processing such as LBP changes, a rich data set is constructed. The wind valve state is divided into three states: full open, full closed, half open and half closed. The data set consists of multiple folders, consisting of the original images and data after the image processing process. One full-state data image is stored for each folder. The authors confirmed the usability of this dataset through previous studies and took data classification methods to obtain some data results. This data can help people studying deep learning and smart subways.