CCTV analytics

CCTV analytics software

The importance of intelligent video surveillance technologies is continuously growing. With a rising number of installed CCTV cameras, manual surveillance becomes an unreal task. Therefore, a security video analysis software that can automatically assess a situation is the most active research field in CCTV today. Due to the high complexity and diversity of CCTV scenes, the current level of artificial intelligence technology still doesn’t allow replacing forensics teams in any way. Even the best deep learning techniques can’t do a complex evaluation of objects’ actions, understand their motives, and make meaningful assumptions. However, an artificial intelligence CCTV software provides a number of valuable video analysis tools that can take over some routines and thus significantly help operators and investigators.

CCTV analytics software converts videostream into information about objects

The full sequence of mental process of a human that studies CCTV video footage looks like this: see a video fragment; convert raw images into a number of objects; extract an object’s properties, such as faces or clothing; analyze the objects’ movements; evaluate interactions between objects; provide a top-level assessment of the situation based on previously revealed details; compare with the situations from his past experience; act appropriately.

A security video analysis software helps in automating the first several steps of this evaluation sequence effectively. The video recognition software based on a deep neural network can reliably detect target objects in raw footage, as well as extract some of their properties. For example, CCTV analytics can track people or vehicles and estimate the object’s colors, movement speed, etc. With such information in hand, a user can already apply the corresponding rules and let the system act automatically. Depending on the type of detected object and applied rules, we’ll get the whole list of intelligent CCTV analytics modules:

  • Vehicle speed detector
  • Unattended object detection
  • Crowd detector
  • Line crossing detector
  • Parking spot detector
  • Traffic rule violation detector
  • Incorrect movement direction detector
  • People loitering detector
  • Fall detector
  • Object proximity detector
  • Unusual motion detector

This list could go on – there are many other detectors that rely on object detection and tracking. Although the object detectors appeared long before the neural networks, only with neural-based detectors and trackers their reliability reached the level of human operators. Today the artificial intelligence CCTV modules can detect people, vehicles, and other arbitrary objects even in a harsh outdoor environment almost as precise as the human brain can.

BitRefine Heads is a multipurpose recognition platform that has a deep neural network “brain” in its core. The user is free to configure the video processing pipeline according to his task. For example, with a few clicks, you can make your system automatically alert you when a person in a red uniform goes in the wrong direction. To do so, you first add a person detector to the recognition pipeline. This neural detector will locate all people in the frame. Then you add a tracker that will take the information about detected people and build a track of their movements. Next, you add the object’s color recognizer that will extract the color of a person’s outfit and save it as a property. After that, you draw a line that will record an object’s movement direction. Finally, you define the circumstances when the system shall notify you. Apart from the user-defined notification, the system saves all events in DB with reference images. This allows you not just to check the fact of your rule violation, but also see statistics.

video analytics software's report

One of the key features of the BitRefine Heads platform is that it can simultaneously play the role of a guard and business intelligence tool. It can collect the statistics of visitors, vehicles or other objects and present the results as interactive business reports. For example, managers can get charts that show how many objects pass into each direction per an arbitrary time interval. Such information may be meaningless for the security personnel, but for business management, it may be of great value. Companies can (and should) use their existing CCTV infrastructure as a source of business insight. Today more people are paying attention to data’s value as an asset. In terms of improving the process efficiency, the proper use of existing data assets can result in two to twenty percent of cost savings.

We’ve got more CCTV analytics to share. Let’s be friends. But you first...