X-AID™ Telegra's Automatic Video Incident Detection

Revolution in video analysis!

X-AID™ Automatic Video Incident Detection (AVID) is different from others because it implements Artificial Intelligence and Machine Learning algorithms to achieve 75% less false alarms ratio and 50% better detection accuracy.

Area of Use:

  • Tunnel AVID and Traffic Data
  • Road/Highway/Bridge AVID and Traffic Data
  • Traffic Violations and Irregularities Detection
  • Traffic Flow Monitoring

Detection:

  • Wrong-way Driving
  • Stopped Vehicle
  • Traffic Slowdown/Congestion
  • Slow vehicle
  • Pedestrians
  • Reduced or Loss of Visibility (smoke, fog, etc.)
  • Debris on the Road
  • Zig-Zag Driving (NEW)
  • Aggressive Driving (NEW)
  • Tailgating (NEW)
  • Truck /Slow Vehicle in High Speed Lane (NEW)
  • Yellow Box Violations (NEW)

Not affected by common problems:

  • Camera Vibrations
  • Weather Conditions
  • Variations in Illumination
  • Low Video Quality

Statistics:

  • Average Vehicle Speed
  • Traffic Volume
  • Vehicle Classification

Automatic Video Incident Detection (AVID)

Automatic Video Incident Detection is the fastest way to detect a traffic incident and, as such, exponentially increases traffic safety.

Unfortunately, AVID was not the most reliable tool. Until now.

The most common issues related to video-based automatic incident detection systems are:

  • Many false alarms, turning into a seriously dangerous “cry wolf” problem (i.e. once an incident really occurs nobody will pay attention).
  • Incidents not recognized by AVID, thus no or slower reaction to that incident.
  • Another standalone subsystem in the control room that requires space, another workstation, extra personnel, etc. when it should be integrated with the rest of the systems in a single system with a single point of control at the control room.

First two issues are closely related. In any traditional automatic video incident detection system there is always a tradeoff between false alarm rate and low detection rate. The question is how to choose lesser of two evils, both of which directly endanger traffic safety. X-AID™ answer is to isolate the most common causes of unreliable operations and to eliminate them. For more information click here).

A typical example of the third issue occurs when the AVID system staff upgrade/correct/calibrate the system, and the Traffic Management can’t see any of the AID alarms on the Traffic Management software or Tunnel SCADA. At the same time, none of the automatic procedures work anymore. The safety in the tunnel is now seriously compromised, and possible consequences are fatal.

Contrary to the example above is Telegra solution - having an AID seamlessly integrated in Telegra’s topXview™ traffic management software, the issue from the above example simply cannot occur. Furthermore, the overgrowing pile of computers and monitors in the control center will be drastically reduced, as there will be no need for separate automatic video incident detection system computers. When AVID is a part of topXview™ system, it is then natively integrated with the video subsystem, as well. That allows for any possible incident to be checked and verified at the very same system, ensuring the fastest reaction to an incident, and thus providing much higher traffic safety.

Additionally, built-in machine learning setup process and automatic calibration ensure that there is no need for periodic setup or calibration. This drastically reduces the need for AVID system maintenance.

Telegra’s X-AID™ AVID can also be used as a standalone system, if needed. This is the beauty of the platform – one can scale it to only automatic video incident detection system, if needed.

X-AID™ Videos

XAIDTM - Superior to Traditional Systems - Learn Why

XAIDTM - Areas of Use

 

Some of the examples on how X-AID™ handles situations that are problematic for traditional video detection:

Direct light influence

Camera vibrations

Luminance variations

Bad weather conditions

Detection of camera shift

Recognition of small objects and complex scenes

Pedestrians detected between vehicles

Static shadows

Moving shadows (that in conventional AID often causes false alarm “wrong direction”)

Overview

Main purpose of Video Automatic Incident Detection System is to increase safety on the road.

With hundreds of cameras deployed throughout highways and tunnels delivering the immense amount of information simultaneously, it is impossible even for a group of human operators to monitor and evaluate the information effectively or efficiently. Reliable Video Automatic Incident Detection system is a must in such environments because it significantly improves the efficiency of traffic management systems. It automatically reports dangerous situations and irregular traffic conditions, such as fire or smoke in a tunnel, stopped vehicle, slow vehicle, driving in opposite direction, pedestrians in traffic lanes, dropped cargo, or traffic congestion. Lack of AID system means not to have an information on time (or not to have it at all!) about various incidents on the road, e.g. collision. These could mean further accidents as a result of queue and/or consequentially other damages.

Main constrains in Video Automatic Incident Detection industry is never ending trade-off between the need to have the least possible number of false alarms and still not to miss some of alarms that did occur on the road.

On one side frequent false alarms could lead to Cry Wolf Syndrome - too many false alarms cause the operator to stop paying attention and not react when there is a real need for reaction. Other side of that medal is that if you do not alarm when something really happened you have endanger the safety of the traffic and have missed the real purpose of the Video based Automatic Incident Detection system.

Today’s conventional AID systems have come to dead end with technology currently being used.

X-AID™, new generation of AID have made a big step further and are solving critical issues conventional AID systems can’t solve - unique and the most advanced algorithms based on cognitive decisions principle open path to the new reliability standard in video detection industry. Machine learning setup process enables fine tuning for the best fit to the particular scene and environmental condition and experience based reliability and accuracy improving process.

Case Studies

Click to open documents:

M4 Russia

Fløyfjell Tunnel

Amir Kabir Tunnel