Real-Time Dog Detection and Alert System using Tensorflow Lite Embedded on Edge Device

Wild and domestic animals constitute a massive part of the animal kingdom. There are approximately 8.7 million different species of animal available. Still, many species are being discovered. Cats and dogs are the most common animal in the society where people live and work. Human beings are also social animals who know society's living norms. However, animals like dogs and cats are not aware of the rules and norms to be followed in society. Hence it is necessary to keep track of the animals in the restricted areas to avoid sudden death, which happens accidentally. So, this paper proposed a real-time dog detection model with email notifications using Embedded ML running on Raspberry Pi for the University campus. Machine learning models like SSD, YOLO v3, and YOLO v4 are used to detect the dog and send the notification to the concerned authority. It has been found that the YOLO v3 model outperforms all other models with the highest accuracy of 94%, whereas SSD and YOLO v4 Achieved an accuracy of 89% and 63%, respectively.