With the rapid proliferation of low-altitude Unmanned Aerial Vehicles (UAVs), unauthorized and disorderly flights have posed persistent threats to public safety, critical infrastructure, and low-altitude airspace management. Stable and reliable detection of small UAVs in complex environments has therefore become a key requirement for anti-UAV systems. Existing reviews are commonly organized according to sensor types or algorithmic architectures, making it difficult to reveal the intrinsic relationships among different sensing modalities. To address this issue, this paper establishes a unified classification framework oriented toward detection criteria from the perspectives of the physical sources and sensing mechanisms of the differential information between UAVs and their backgrounds. Accordingly, anti-UAV target detection is categorized into three types: static appearance differences, dynamic temporal differences, and intrinsic radio-frequency differences. Within this framework, the fundamental principles, representative methods, development trends, advantages, and limitations of single-modal detection technologies, including visible-light imaging, infrared imaging, radar, acoustics, and radio-frequency detection, are systematically reviewed. Multi-modal detection methods are further analyzed in terms of multi-source collaborative modeling, cross-modal alignment and fusion strategies, together with the engineering applications of representative anti-UAV detection systems. In addition, publicly available datasets are summarized according to sensing modality, and commonly used evaluation metrics for different modalities, including detection accuracy, localization accuracy, real-time performance, detection range, and environmental robustness, are reviewed. Finally, future research directions are discussed in light of current technical bottlenecks, including multi-modal collaborative sensing, cross-scenario generalization, robust feature learning, and lightweight deployment.

With the development of artificial intelligence technology, Unmanned Aerial Vehicles (UAVs) can automatically identify and locate objects of interest by equipping different optoelectronic payloads, and are used for Search and Rescue (SaR) missions in disaster environments. UAV search and rescue missions require the rapid location of missing persons in complex and dangerous environments, facing challenges such as single detection technique, susceptibility to interference during imaging, complex background, and limited UAV platform resources. In recent years, the integration of UAV systems with deep learning algorithms has emerged as a prominent research focus. To demonstrate the advancement of UAV search and rescue techniques, the review of these methodologies has been conducted. Firstly, the difficulties faced by UAV search and rescue are analyzed according to the characteristics of UAV search and rescue missions. Subsequently, various UAV search and rescue methods based on deep learning in recent years are sorted out. Additionally, object detection datasets relevant to UAV search and rescue missions have been sorted out according to different data types. Finally, the problems of current research and future development directions are summarized.

In the context of globalization, the importance of ship monitoring is becoming more and more prominent. With the continuous progress of the remote sensing imaging technology, ship detection has become a key means to ensure the safety and efficiency of maritime transportation, and is crucial for maritime transportation, environmental protection, and national security. However, due to the large difference in scales and complex background of ship targets, existing single detection model methods rely too heavily on training data and cannot adapt to ship targets with variable scales. In this paper, we propose a multiple models collaboration framework, in which multiple trained ship detection models are regarded as auxiliary network, and the main network training is optimized by knowledge migration. First, ternary relationship constraints are introduced to transfer knowledge between the auxiliary network and the main network. Then, a soft-label guidance strategy is proposed to further improve the accuracy of ship detection. The experimental results show that compared with the existing mainstream methods, the proposed method demonstrates better performance on DOTA and xView datasets, overcoming the limitation of a single model and providing a new solution idea for target detection in remote sensing images.