Research on A Pest Inspection System for Small Mountain Orchards Based on Image Data Analysis

China’s total fruit planting area and output have always been in the forefront of the world. Fruit has become the largest planting industry. Therefore, in order to reduce economic losses and environmental pollution, obtain healthier and more productive fruits, and promote the continued development of the fruit industry, one of the problems that must be solved is to better control pests. Many orchards in China are planted in remote mountainous areas, with inconvenient transportation. It is difficult to find pests and collect relevant data because the site can not be guarded for a long time. At present, many smart farm devices in the market have the disadvantages of large volume, complex installation process and high cost of data acquisition sensors. Based on image data analysis technology, combined with PyTorch deep learning framework, OpenMV module and unmanned aerial vehicle, this paper designs a pest inspection system for small and medium-sized orchards suitable for planting in mountainous areas. The system is divided into PC side model training and embedded side model deployment, including software apps and hardware devices, which are easy to operate, small in size and low in cost. Its functions include pest identification, environmental data detection, image transmission, crop dynamics, orchard pest control knowledge learning manual, farmers’ forum, marking hill fires and other modules, aiming to provide the most suitable services for mountain fruit farmers.