Comparing Six Computer Vision Techniques for Rice Pest Recognition in the Field

The brown planthopper (BPH) is an agricultural pest that causes up to ten percent reduction in annual rice crop yields across a wide range of countries, from Pakistan to Japan. Identifying agricultural pests using computer vision techniques has recently gained attention as a research area. However, most studies evaluate these pests under artificial conditions, which limits their external validity. In this paper, we utilize six computer vision algorithms (BRIEF, HOG, LBP, SIFT, SURF, and ORB) to identify BPH from field photos. We evaluate each algorithm’s precision, recall, and accuracy and aim to demonstrate the feasibility and effectiveness of the best methods for designing, implementing, and applying pest identification.