As small‑sized agricultural pests with wide‑ranging hazards, rice planthoppers are difficult to monitor accurately by traditional manual inspection, which suffers from low efficiency, large measurement error and obvious time‑lag and fails to satisfy the requirements of precise pest prevention and control. This paper reviews real‑time counting and localization technologies for rice planthoppers based on the Fully Convolutional Network (FCN). Its core mechanism extracts multi‑scale features of rice planthoppers through a lightweight backbone network. Combined with the channel‑spatial hybrid attention mechanism and Feature Pyramid Network structure, the recognition performance for tiny targets is enhanced. Pixel‑level localization and blob prediction of rice planthoppers are realized by adopting object‑oriented loss‑function, so as to complete accurate counting. Breaking through the detection bottleneck caused by the tiny size, frequent overlapping and complex background of rice planthoppers, this technology achieves real‑time response under field conditions. It provides an efficient tool for dynamic monitoring of rice‑planthopper populations, supports precise pesticide application and green pest‑control strategies, reduces the risk of pesticide residues, improves agricultural productivity and sustainability, and supplies technical support for the development of intelligent plant‑protection and precision agriculture.