Based on the image recognition training of large-scale apple crop pests and diseases, it automatically recognizes the types and extents of pests and diseases in apple crop pictures, with an accuracy rate of over 90%, and timely defense, early treatment.