In order to enable a quick and effective initial response to domestic pests, we intend to establish high-quality learning data that can detect and classify automated pests using artificial intelligence technology and open it to lay the foundation for establishing a system to minimize damage to farms caused by pests.
It is active from July to November and collects 11 kinds of pests that damage crops, and installs a collector consisting of sticky, fermon, and infant types in the pest area to collect 120,000 image data with the built-in camera. In addition, 140,000 environmental sensor data are collected with installed sensors and processed into data that AI can learn.
Finally, Detectron2 based on Mask R-CNN model and Random Forest models are used to build an object detection classification model and a growth environment analysis model through AI deep learning.