Deep learning and image processing provide flexible techniques for human activity detection. This research presents a fresh perspective, since it is difficult to determine individual worker pay directly using pictures and machine learning (ML). We observe in our daily lives that image processing has advanced to a good level in a number of disciplines, including security, law enforcement, medical imaging, and visualization. Through task-specific action analysis, the idea extends beyond simple labor count calculations. The work that is currently being done consists only of operations that use strategically positioned cameras to take pictures of workers in factories. Lastly, by improving accuracy, the errors are examined and the system is adjusted to produce better outcomes. However, there’s a twist: The system can infer task involvement by analyzing worker actions; this indirectly relates tasks to wages based on predefined wage structures. The model is trained on films exhibiting diverse task-specific activities. By removing the “limited image relevance” barrier, this technique allows for the extraction of more information from activities than just postures and facial expressions. But ethical issues like transparency and privacy are still very important. The goal of this research is to create computer techniques for activity identification in commercial factory labor wage computation.