Tom Cavey and Jason Stock Helming from Colorado State University have reportedly published a paper on a new Artificial Intelligence system that rewards dogs upon listening to voice commands and completing tricks. Cavey and Stock are Computer Science students who trained cutting-edge image classification networks to regulate if a dog is standing, sitting, or even lying down. When any furry canine responds to a command, the machine powered by AI dispenses a treat.
The grad students used Nvidia’s next-gen autonomous machine, the Jetson edge platform, for recognizing tricks and dispense threats in real-time. Nvidia Jetson™ is capable enough to deliver both power efficiency and performance for running AI-powered autonomous machines faster, using optimum power. Cavey and Stock’s AI machine’s real life application could become a mainstream aid for dog trainers globally. It is important to note that the machine schools dogs’ innate behavior in households. Stock released a statement suggesting that they have paved the next-gen dog treat dispenser, which can be a product in the foreseeable future.
Dog training data assisted in hassle-free treat dispensing
Initially, the researchers came up with dog images exhibiting three distinctive yes specific postures. The AI had to learn from over 20,000 images comprising several sizes that depicted various positions in dogs. The images were further preprocessed, so they had to write a program to help them label dog’s positions seamlessly.
On the other hand, Nvidia came forward to reveal that the AI machine doesn’t work remotely as of now. With time, the AI dog dispenser will be modified into a robust remote machine. The device also uses the software development kit from Nvidia, which powers drones and several other systems.