Implementation of Temporal Action Detection Workflow for Video Data Analysis with the Use of Machine Learning Operations

The article aims to discuss the peculiarities of using the Machine Learning Operations paradigm in the context of addressing the task of temporal action detection in video data. The results of applying Machine Learning Operations to the aforementioned task are analyzed at three levels of machine learning workflow, namely, at the level of machine learning system design, at the level of machine learning system components, and at the level of elements within the machine learning system components. It is emphasized that Machine Learning Operations facilitates the procedures of designing, developing and deploying the temporal action detection workflow.