Abstract: Rice is the main grain crop in China, and pests are one of the key factors restricting its high quality and yield, imposing huge control pressure on growers. At present, rice pest detection still relies primarily on manual investigation and sensors, which have inherent defects such as low efficiency and poor real-time performance. These shortcomings make it difficult to timely detect pests at their early stages of occurrence, leading to the problem of extensive pesticide application, and causing economic losses and ecological damage. Therefore, conducting rapid and accurate intelligent detection of rice pests is of great significance for ensuring food security and promoting the development of green agriculture. With the iteration of information technology, intelligent detection technology for rice pests has made significant progress. This article reviews technologies for rice pest detection and their respective advantages, analyzes the internal constraints of various technologies, and then predicts the development trends of intelligent pest detection technology. Among them, sensor technologies such as infrared optoelectronics, acoustics, electronic nose, and insect hormones are susceptible to interference from complex field environments and have insufficient stability in practical applications. Although traditional machine learning-based rice pest detection method has certain accuracy and application foundation, the results are easily constrained by subjective factors and have limited adaptability due to the reliance on manual feature extraction. The advantages of rice pest detection methods based on deep learning algorithms are prominent, not only with high recognition accuracy, but also with significantly reduced labor costs. Although there are certain requirements for experimental conditions and data volume, these can be gradually met through optimization at the technical and data levels. The fusion scheme of lightweight neural networks and edge devices has achieved a dual breakthrough in real-time detection and convenience, providing strong support for practical implementation in field scenes and having broad application prospects and promotion value. This review considers that deep learning is still the forefront technology direction in the field of intelligent detection of rice pests. Specifically, at the application level, in the future, it is necessary to improve the detection accuracy of models in actual field environments such as low light and shadow occlusion through multi-scale prediction optimization and feature extraction network enhancement. At the same time, it is necessary to further streamline the number of model parameters and computational complexity, improve recognition speed while ensuring accuracy, and combine IoT technology to achieve adaptation with edge devices, laying the foundation for field deployment and practical applications.