© 2024 AtFrame
Video archive and video broadcast management system
The video archive and video broadcast management system is designed to create an archive of video recordings and broadcast them in FullHD and 4K quality.
To form a television-quality image, it is proposed to use cameras with a 20 Mpix matrix with a PCIEx interface, which will allow dynamically forming the working area of the image,
A solution based on XIMEA cameras (Slovakia) in combination with NVIDIA Jetson Xavier AGX Industrial calculators is offered as a hardware platform. It is proposed to ensure the formation of close-ups by using cameras with PTZ control and a focal length of the lens of 23-750mm or 23-300 mm (the exact choice will be made after final configuration approval based on the results of the first stage of work).
The video archive is supposed to be stored on the basis of a storage system –a tape library based on LTO-9 cartridges, which provides a volume of more than 9 Pb per rack at a relatively low cost of ownership.
A 120 Tb HDD drive and a 20 Tb fast buffer based on SSD drives will be used as a buffer.
The video archive and video broadcast management system is designed to create an archive of video recordings and broadcast them in FullHD and 4K quality.
To form a television-quality image, it is proposed to use cameras with a 20 Mpix matrix with a PCIEx interface, which will allow dynamically forming the working area of the image,
A solution based on XIMEA cameras (Slovakia) in combination with NVIDIA Jetson Xavier AGX Industrial calculators is offered as a hardware platform. It is proposed to ensure the formation of close-ups by using cameras with PTZ control and a focal length of the lens of 23-750mm or 23-300 mm (the exact choice will be made after final configuration approval based on the results of the first stage of work).
The video archive is supposed to be stored on the basis of a storage system –a tape library based on LTO-9 cartridges, which provides a volume of more than 9 Pb per rack at a relatively low cost of ownership.
A 120 Tb HDD drive and a 20 Tb fast buffer based on SSD drives will be used as a buffer.
The system for soccer video analytics
The system of soccer video analytics is based on the analysis of video recordings of games and training sessions. The solution will be based on the existing image analysis library based on NVIDIA GPU of its own design and the Fastvideo library for image rescaling and RAW>>RGB conversion.
The use of calculators attached to the cameras will provide the ability to convert RAW>>RGB and generate a stream of JPEG images. This will ensure the organization of fast replay (this task is not solved by standard H264 codecs) and the prompt selection of ROI for broadcast. In this case, the search for players and the ball will be performed on a reduced resolution image, which will provide the required performance.
1. Monte-Carlo based 2D object tracking approach in high load scenes Gilya-Zetinov, A.A.,Tsybulko, E., Bugaev, A., Zaitseva, A.S., Pashkov, R. // Proceedings, International Conference on Electrical, Control and Instrumentation Engineering, ICECIE, 2022-November
2. High-Speed Multi-person Tracking Method Using Bipartite Matching Gilya-Zetinov, A., Bugaev, A., Khelvas, A., Konyagin, E., Segre, J. //Lecture Notes in Networks and Systemsthis, 2022, 283, pp. 793–809
The system of soccer video analytics is based on the analysis of video recordings of games and training sessions. The solution will be based on the existing image analysis library based on NVIDIA GPU of its own design and the Fastvideo library for image rescaling and RAW>>RGB conversion.
The use of calculators attached to the cameras will provide the ability to convert RAW>>RGB and generate a stream of JPEG images. This will ensure the organization of fast replay (this task is not solved by standard H264 codecs) and the prompt selection of ROI for broadcast. In this case, the search for players and the ball will be performed on a reduced resolution image, which will provide the required performance.
1. Monte-Carlo based 2D object tracking approach in high load scenes Gilya-Zetinov, A.A.,Tsybulko, E., Bugaev, A., Zaitseva, A.S., Pashkov, R. // Proceedings, International Conference on Electrical, Control and Instrumentation Engineering, ICECIE, 2022-November
2. High-Speed Multi-person Tracking Method Using Bipartite Matching Gilya-Zetinov, A., Bugaev, A., Khelvas, A., Konyagin, E., Segre, J. //Lecture Notes in Networks and Systemsthis, 2022, 283, pp. 793–809
AI based video recording for sport and entertainment
AI solution for selfies in entertainment and sport industry
Main scenario for system is:
Main scenario for system is:
- The athlet reads the QR code at the entrance and after that the cameras track this athlet for 1-5 minutes
- Video is transferred to the cloud service and processed for Tik-Tok/YouTube etc
- The processed video is posted on athlets’s social networks
- Video uploads to the storage and CDN
Communication system integrated with AI
We propose the TETRA RF network integration with IP and Cell Phones communication channels
Our advantages:
We propose the TETRA RF network integration with IP and Cell Phones communication channels
Our advantages:
- Gates to 4G and 5G solutions
- Gates to IP network
- AI agents for quick assistance
- Voice based commands for Machine Vision and Robotics solutions
The Innovations:
AI abonent in Tetra RF
