University of Bristol researchers have developed a non-intrusive visual surveillance system for wildlife habitats. They've used their system to monitor the behavior of 20,000 African penguins on Robben Island in South Africa. By definition, conventional tagging techniques can only monitor animals which have been tagged. On the contrary, the 'Penguin Recognition Project' relies on visual recognition software. The scientists claim that they can correctly identify an individual penguin 'with around 98 per cent reliability' -- how can they measure this? They also claim that their approach could be used to monitor other endangered species, such as zebras or sharks. But read more...
You can see above this "computer vision system in action identifying African penguins on their way from the colony to the beach. Green boxes indicate the penguins detected as species members in near-frontal poses, a yellow bounding box shows that a penguin has been identified as an individual." (Credit: Spot the Penguin Project)
The researchers have taken advantage of a specific characteristic of these African penguins. "African penguins carry a pattern of black spots on their chests that does not change from season to season during their adult life. As far as scientists can tell, no two penguins have exactly the same pattern. The researchers have developed a real-time system that can locate African penguins whose chests are visible within video sequences or still images. An extraction of the chest spot pattern allows the generation of a unique biometrical identifier for each penguin. These biometric data can then be used to identify individual, African penguins from video or photographic images by comparison with a population database."
They also claim that their recognition system works very well and could be used to monitor other endangered species. "Provided that a good image of a penguin can be extracted, the system can correctly identify the individual with around 98 per cent reliability. The current limitation of the system, based on one camera, is that some passing penguins are hidden behind others, or the lighting is poor. The researchers are currently working to overcome these limitations both by combining images from intelligent pan-tilt-zoom cameras, and by using infra-red imaging to provide data both day and night. The basic image-recognition system has also been trialled with zebras, sharks and, in principle, can be extended to any species with complex surface patterns."
You'll find more information by visiting the Spot the Penguin Project website. This project description will give you access to a 19 seconds video from which I've picked the above picture. It also gives more details about the current prototype of the project.
- Vision Software: As one essential part of our research we focus on developing the 'intelligent' software that allows systems to make sense of complex camera images and interpret animals and their patterns as individual entities.
- Hardware Architecture: Our current prototype system has a distributed design: client systems gather data at different locations in the penguin colony while a central server holds the population data. All necessary software components are created to run on relatively inexpensive consumer hardware.
- Data Flow and Networking: The cameras capture images and send a time-stamped version of them in a live stream to locally connected laptop computers which identify members of species in real-time. The relevant areas of interest in each image (the penguin chest patterns) are then transferred to local hotspots using a wireless network.
Finally, you can participate in a competition named Can you spot the penguin? Here are the simple rules. "If you can do what our Penguin Recognition System can do, you have the chance of winning a trip to work as a volunteer on an Earthwatch expedition for two weeks with the penguins on Robben Island in South Africa."
Sources: University of Bristol news release, June 27, 2008; and various websites
You'll find related stories by following the links below.