The safety risk of underground operation is prominent, the object of supervision is many, and the situation of safety control is severe. The actual oilfield operation types are complex and diverse, contractor personnel mobility, personnel quality is uneven, there are many supervision objects, security risk control difficulties. Traditional video surveillance requires manual analysis and judgment, and early warning linkage is insufficient. Therefore, intelligent management and control, information fusion and deep application of data are the core links of intelligent oilfield construction and safe production. Intelligent well site monitoring includes intelligent monitoring of well site data and intelligent control of operation video. The research group proposed an intelligent well site monitoring method combining machine learning algorithms with well site job types. Based on big data analysis, it conducted job supervision, inspection and driller behavior evaluation, and used target detection algorithm and posture recognition method to realize personnel violation monitoring and operation fault warning. By analyzing the key points of safety production risk control in wellsite operation, it enrichis on-site monitoring means, intelligent video analysis, information fusion and deep application of data, and realizes intelligent monitoring, intelligent alarm, intelligent operation, intelligent supervision and intelligent management with human brain functions from single video monitoring "seeing" to multi-dimensional "seeing, hearing, hearing and speaking". Finally, well site operations are fully controlled.