The cameras give Gabon wildlife rangers a new tool in the fight against poaching by capturing pictures of trespassers. The systems can also monitor biodiversity loss by counting the number of animals in an area. “Regular cameras can activate ‘mechanically’ when something triggers them, for example, movement or sound,” AI expert James Caton told Lifewire in an email interview. “AI embedded in the camera can more intelligently activate when items of interest pass within the frame – for example, a person or poacher versus a moose. AI can differentiate between human figures and animal figures, for example, by posture or size.”
Computing on the Edge
Thanks to AI, the new camera traps developed by the group Hack the Planet are more intelligent than previous models. The system uses machine learning to analyze photos in real-time on the device to detect animals and humans. The traps alert rangers if an elephant, rhino, or human movement is detected. Equipped with a satellite uplink, the system can operate anywhere globally without depending on a GSM or Wifi network. Stirling University researcher Robin Whytock and a team of researchers tested an AI model to analyze camera trap data. The case study they used classified Central African forest mammalian and avian species. And even with a relatively small dataset of 300,000 images used to train the model, the outcome was strong, the researchers reported in a paper. The researchers said the machine algorithm was 90 percent accurate and can classify about 4,000 images per hour on desktop machines used by park rangers and ecologists in the field, without access to powerful cloud computing resources. The AI system reduces the time needed to analyze thousands of trap images from several weeks to a single day.
Guarding the Trails
Another system called TrailGuard AI is used as a security system for national parks to detect, stop, and arrest poachers. The technology helps improve intelligence on poaching and related illicit networks, helping authorities crack down on the illegal wildlife trade. Small enough to conceal along trails, TrailGuard AI’s camera head uses artificial intelligence to detect humans within the images and relays pictures containing humans back to park headquarters via GSM, long-range radio, or satellite networks. The TrailGuard AI technology was field-tested in a reserve in East Africa, where it helped in the arrest of thirty poachers and the seizure of over 1,300 lb. of bushmeat. Conservationists benefit from AI running in the camera rather than in the cloud because the biggest drain on battery life is not running inference on a computer vision chip in the camera, but the transmission of the image over GSM or satellite modem, Eric Dinerstein, the director of WildTech at the wildlife conservation group RESOLVE told Lifewire via email. Dinerstein said the system accurately weeds out false positives when a camera is activated by something other than a poacher. “In our deployments of TrailGuard in the field, up to 95% of the triggers of the motion sensor are the result of false triggers or false positives,” Dinerstein added. “Only 5% are actual poachers.” TrailGuard can save battery life. Transmitting thousands of false-positive pictures over the course of several weeks runs down batteries. By filtering out the false positives on the edge and only transmitting true positives or very few false positives, batteries can last years. “Also, the chip we use is very low power, and our device is in sleep or power-off mode for most of its life,” Dinerstein said. “Battery life for sensors in remote areas is critical.” Wildlife monitoring could soon get even smarter. Researchers are working on programmable AI embedded in cameras. Currently, images must be retrieved from a camera and processed in the cloud. But new capabilities allow for users to create customized AI agents and deploy them on cameras. “For poachers, for example, if you know they travel in a white car or one of them always wears a yellow cap, you could potentially update the cameras from afar with this new information,” Caton said.
