Video demonstrations

Playing the videos requires an HTML5 conformant browser. The most recent versions of all common browsers will work. If the video does not appear please check that your browser is set to allow the scripts and ActiveX controls that are needed to display video content.

The key factor distinguishing Cobalt from other video detection systems is that it attempts to store all relevant data about a scene along with a compressed version of the scene video. What is meant by "relevant" in this context is "any change of significance that might potentially be the source of a video query". So it tries,for example, to ignore changes in illumination, persistent foreground or background interference, or moderate camera shake.

The approach that Cobalt takes is one of solving a specific problem relating to a video scene while at the same time

One of the main issues with most outdoor video detection systems is an inability to react to and automatically cope with diverse weather conditions. This demonstration shows Cobalt dealing with a camera looking through a window during heavy rain. By the same token, the same setup can be used through day and night and through variations in sunlight.
The simplest setup for Cobalt event detection merely involves drawing boxes around areas of interest in the scene. Any significant change in those areas will be flagged as an event. There are no parameters to set although the way the detection is made can be changed in the so-called "expert mode".
This example shows hours of recording being searched in a matter of seconds. The material is an archival videorecording spanning many days. The program time has to match the reported on-screen video time.
The "expert mode" allows complex video queries to be formulated graphically. This is the first of 6 video lessons demonstrating the variety of queries that can be formulated, from directional motion to discriminating differnt sources of activation. A number of different scenarios are listed
The tracking of flying objects, in this case birds, is an ever more important aspect of video detection. Here we mark the trajectories of three birds in real time with low latency. The positions of the birds can be sent to another suitable device. This was developed to monitor White Eagles hovering over large land-based wind farms.
A multi-camera GUI for managing multiple cameras. Various preset image layouts can be user configured and selected from the buttons at the bottom of the screen. The videos are displayed in the main panel. The left panel is a camera directory that groups cameras into specific areas of the site. Video can be redirected to other monitors and shared by other authorised users.
Complex backgrounds are a frequent source of confusion in video event detection. Here we see the detection of flying seagulls against a background of complex sea-surf movements. This works in part because the Image-DNA tracks textures: the birds are disturbing the background texture.
Complex foregrounds are a frequent source of confusion in video event detection. Here we see the detection of people moving behind a foreground of swirling smoke from a fires. This works in part because the Image-DNA tracks textures: the people have a different texture than the smoke.