AI/ML: video stream tagging

Build a model of automatic tagging and analysis of video files to track the analysis for their further processing.

Implemented by us, the system uses state-of-the-art machine learning models based on convolutional neural networks to speed up the learning process and forecasting with GPU. It supports both lighter and heavier models and has a module for dataset preparation. The module allows for parallel dataset markup and a semi-automatic mode utilizing existing models.

Our multi-node scalable platform uses hardware GPUs for automatic video analysis, with machine learning models and object tracking. It has a user interface for video library maintenance, viewing and uploading, and model selection. A video markup subsystem was added to generate new data sets for model training.