An Agricultural AIoT Bird Repellent System with Machine-Learning based Moving Object Detection

Guan-Hsiung Liaw, Chia-He Li

AIoT (Artificial Intelligence of Things), Bird Repellent System, Deep Learning, YOLO (You Only Look Once).

In recent years, with the maturity of deep learning technology, the problem of image recognition has become easier to solve, and it has begun to be applied to various industries. We try to apply this technology to solve the problem of bird damage in traditional agriculture, because in addition to natural disasters, the largest case of agricultural damage should be the gnawing disaster of birds or insects. In this paper, we propose an artificial intelligence of things (AIoT) bird repellent system architecture and make its prototype. In this system, we trained a neural network model that can identify bird flocks, and developed moving bird flock identification technology based on this, which will actually be implemented on a small single-board computer equipped with GPU (such as NVIDIA Jetson Nano) . The system can be set up in farmland to capture real-time video through a camera and identify the moving bird flocks by a small single-board computer equipped with GPU. When a flock of birds appears, the small single-board computer transmits a message through the LoRa low-power long-distance wireless communication interface to the ultrasonic bird repellent devices deployed in the farmland, and emits ultrasonic sound to drive the bird flock away.

https://doi.org/10.31871/IJNTR.7.12.13