The performance of the human factor will continue to have a significant impact on the overall manufacturing system efficiency, which means that this performance will need to be monitored and quantified. Time-based metrics are viewed as one of the important performance indicators. Conventional approaches to the time study are based on observation of human activities, predefined time standards, and more recently videos and manual analysis of videos. The deficiencies of existing approaches to the time study and technological progress are the motivation for the research of artificial intelligence techniques with the aim of automating action detection in manufacturing processes. Analysis of human activities from a video is one of the research problems in the domain of computer vision, which belongs to the field of artificial intelligence. Time study requires recognition of the activity performed by the subject and the duration of that activity. The simultaneous solution of these two tasks represents one of the research directions in the field of computer vision, which in the context of production is still not sufficiently investigated. Modern approaches to solving the problem of simultaneous recognition and time segmentation of activities are based on deep learning, an area from the domain of artificial intelligence. The goal of this research was the development of a deep learning model with the capability of recognition and temporal segmentation of a series of human activities from videos collected in manufacturing processes. To achieve this goal, a sample was collected from the real manufacturing process, which consists of nine work activities. During the video recording of the process, the work activities were performed by four subjects on three different types of products, while the recording itself was performed from two different view positions. 27 different models have been developed which differ with respect to recording viewpoint, model input features, and model architecture responsible for activity classification and time segmentation. In the dissertation, a procedure for evaluating efficiency and comparison of new models is proposed. The developed models will contribute to the work of industrial engineers, facilitating the analysis of human factor productivity in manufacturing processes.