IDIS AI in the Box (DV-2116);
Accurate, fast and effective
IDIS introduces AI in the Box (DV-2116), reinforcing the power of video observation systems with the most accurate Deep Learning Analytics. IDIS Deep Learning Analytics (IDLA) has achieved unprecedented accuracy of 98% in independent tests, a record performance with fast data processing.
IDIS DEEP LEARNING
The DV-2116 makes Deep Learning Analytics accessible for small to medium-sized applications, thereby improving the security and efficiency of the control room. The plug-and-play IDLA-ready device comes with an NVDIA GTX1060 GPU chipset that can be used to analyze up to 16 channels simultaneously.
Users benefit from:
- robust and calibration-free object detection and classification (objects such as people, cars and bicycles)
- detection of intruders and hanging around
- powerful intelligent search functions
- follow on color, object and number
The Deep Learning Engine from IDIS, which drives the new DV-2116 AI in the Box solution, can recognize potentially significant movements and characteristics of people and vehicles, while ignoring activities that are not relevant. The technology can quickly go through hours of video to find specific people. Due to the self-learning properties, it also becomes more accurate over time.
IDIS Deep Learning Analytics (IDLA)
An advanced AI engine designed for security and safety applications
The IDIS Deep Learning Engine is an artificial neural network that acts like the human brain, learning from experience. Out-performing human capability it can analyze vast amounts of data points taken from video footage across single or multiple cameras, covering an area, a site or multiple facilities simultaneously. This allows it to first detect and then identify an event or threat, while filtering out false alarms.
The industry’s most advanced and accurate AI engine
The IDIS Deep Learning Engine has been designed and developed in-house by our own team of AI experts using deep neural network technology to deliver industry-leading accuracy. Its capability has been proven in independent testing at live customer sites using high-quality video footage.
Learning from Big Data in Video
The IDIS Deep Learning Engine is constantly improving, with ongoing system development and big data learning – an evolution that started with the first generation in 2017 and has already progressed to our third-generation V3 engine, launched in 2019.
Today, the IDIS Deep Learning Engine is achieving 98% accuracy and outperforming other top-tier surveillance vendors.
Up to 98% accurate
IDIS has developed its Deep Learning Engine specifically for security and life safety applications.
The engine has evolved using big data from surveillance footage where security and safety breaches have occurred – events such as intrusions or people having slips, trips or falls. It has also learned by analyzing data from video streams that capture the behavior of people and the movement of vehicles and objects.
IDIS Deep Learning Analytics allows users to simply and easily leverage the power of the IDIS AI engine.
IDIS Object Detection technology allows operators to automatically identify and track target objects by using modelling technology that creates a fixed background in the scene, allowing the detection and monitoring of moving objects across the top.
In practice it can analyze loitering and trespassing, and handle people counting in both real-time video streaming as well as recorded footage.
IMF (Instant Meta Filtering)
As surveillance footage is recorded, the IDIS AI engine automatically recognizes objects, places, and movements and then extracts and stores metadata relating to every scene. This metadata provides classification, identity and context to video streams, allowing operators to organize, search and retrieve intelligent information from huge amounts of video footage quickly and easily.
IMF rapidly sorts through fine-grained meta-data – as easily as performing a simple text search – to locate a specific person or vehicle of interest across a site or even multiple facilities. In turn this reduces investigation time for often critical incidents from days and hours down to minutes.
Action Recognition (option)
Action Recognition is an AI technology that uses advanced automated algorithms to provide high accuracy and automated event recognition, self-learned from various cases.
For example, Action Recognition can spot a person falling over, in real-time, and trigger a notification. This ensures that critical events never go undetected and enables control room operators to initiate fast, appropriate responses to incidents.
Person Match (option)
IDIS Person Match technology that extracts the characteristics of a person in order to search for the same person or persons across single or multiple recorded video streams.
This allows users to quickly search vast amounts of video data to monitor and track people of interest, to reveal their behavior and movements over time, and to allow rapid investigation of incidents or suspicious activity.
Scalable, multi-format solution
Accurate, fast, simple, and scalable, IDIS Deep Learning Analytics (IDLA) powered by the IDIS Deep Learning Engine, offers agile object detection and classification of people, cars, and bicycles, intrusion detection, and loitering detection—all adapted to fit a 16:9 ratio.
IDLA – An AI solution optimized for video surveillance