AIVision - video analytics platform

/ /
AIVision - video analytics platform
Recent advancements in AI-based video analytics have brought remarkable progress in complex fields like object detection, activity recognition, and anomaly detection. With the market poised to grow at a Compound Annual Growth Rate (CAGR) of over 30% in the next five years, this area of technology is experiencing rapid expansion. AIVision, who joined Oranged Fab in 2023, stands at the forefront of these developments as an advanced AI-based video analytics platform, being capable of recognizing of hundreds of object classes, LPR, OCR directly from live security or USB camera feeds, as well as from saved video files.

In a nutshell

In a nutshell, AIVision platform is a comprehensive, AI-driven video analytics system that processes and analyzes video streams for a variety of purposes, including security monitoring, operational efficiency, and data analysis, which make it a powerful tool for organizations across industries. Using the platform, security firms can monitor over 1000 streams with only 2500 euro hardware and receive alarms extremely fast to dispatch in less than 40ms, outdoor advertising companies – measure audiences, and municipalities – identify traffic congestion in real time.Quality inspection, traceability and quality control become processes that can be automated by identifying errors/defects in production workflows using AIVision technology.


Compared to competitors, the platform stands out with a favorable combination of the following features:
  1. Security at its core, the platform utilizes an advanced AI framework supported by a secure, proprietary client-server and encrypted peer-to-peer (P2P) communication protocol. This setup ensures that video data and communications are secure and efficient, facilitating the real-time processing of video streams.
  2. Integration capabilities the platform is designed to be highly adaptable, supporting integration with a wide range of devices from IoT gadgets to virtual processing servers.
  3. Labels management module this is a key feature of the platform, which lets users manage alarm images and categorize them within their account. This module connects to all video streams processed by the client's devices, enabling the customization of object detection classes and the improvement of detection accuracy over time.
  4. Real-time video processing users can add any video streams to the platform for processing. The platform’s AI can analyze these streams in real time, identifying objects, activities, or anomalies based on the preferences applied by a user.
  5. Advanced AI analysis techniques the platform supports Multi-Inference (using different NNs for different areas of a video stream), Inference in Inference (further analysis of detected objects with different NNs), and Contextual Inference (using text-based commands for analysis).
  6. Flexibility and adaptability users can tailor the platform according to his needs by training the system to recognize new object classes or refine the detection of existing ones.
  7. Fast processing by utilizing edge computing and private 5G networks (eMBB slice), the platform ensures fast processing speeds and reduced latency.

Discussing the position of AIVision on market, CEO Bogdan Grumezescu highlights that the integration of processing devices, the communication protocol, and the hardware-agnostic framework solidifies the solution's strong stance, not only in the Romanian AI video analytics market, which is moving towards edge processing. AIVision, supported by Orange infrastructure, is excited to continue advancing and maintaining a leading position in the video analytics market!