Federated learning has received a great deal of research attention recently,with privacy protection becoming a key factor in the development of artificial intelligence. Federated learning is a special kind of distributed learning framework, which allows multiple users to participate in model training while ensuring that their privacy is not compromised; however, this paradigm is still vulnerable to security and privacy threats from various attackers. This paper focuses on the security and privacy threats related to federated learning. First, we analyse the current research and development status of federated learning through use of the CiteSpace literature search tool. Next, we describe the basic concepts and threat models, and then analyse the security and privacy vulnerabilities within current federated learning architectures. Finally, the directions of development in this area are further discussed in the context of current advanced defence solutions, for which we provide a summary and comparison.

At present, 3D reconstruction technology is being gradually applied to underwater scenes and has become a hot research direction that is vital to human ocean exploration and development. Due to the rapid development of computer vision in recent years, optical image 3D reconstruction has become the mainstream method. Therefore, this paper focuses on optical image 3D reconstruction methods in the underwater environment. However, due to the wide application of sonar in underwater 3D reconstruction, this paper also introduces and summarizes the underwater 3D reconstruction based on acoustic image and optical–acoustic image fusion methods. First, this paper uses the Citespace software to visually analyze the existing literature of underwater images and intuitively analyze the hotspots and key research directions in this field. Second, the particularity of underwater environments compared with conventional systems is introduced. Two scientific problems are emphasized by engineering problems encountered in optical image reconstruction: underwater image degradation and the calibration of underwater cameras. Then, in the main part of this paper, we focus on the underwater 3D reconstruction methods based on optical images, acoustic images and optical–acoustic image fusion, reviewing the literature and classifying the existing solutions. Finally, potential advancements in this field in the future are considered.

Temporal Action Detection (TAD) aims to accurately capture each action interval in an untrimmed video and to understand human actions. This paper comprehensively surveys the state-of-the-art techniques and models used for TAD task. Firstly, it conducts comprehensive research on this field through Citespace and comprehensively introduce relevant dataset. Secondly, it summarizes three types of methods, i.e., anchor-based, boundary-based, and query-based, from the design method level. Thirdly, it summarizes three types of supervised learning methods from the level of learning methods, i.e., fully supervised, weakly supervised, and unsupervised. Finally, this paper explores the current problems, and proposes prospects in TAD task.

In action recognition, obtaining skeleton data from human poses is valuable. This process can help eliminate negative effects of environmental noise, including changes in background and lighting conditions. Although GCN can learn unique action features, it fails to fully utilize the prior knowledge of human body structure and the coordination relations between limbs. To address these issues, this paper proposes a Multi-level Topological Channel Attention Network algorithm: Firstly, the Multi-level Topology and Channel Attention Module incorporates prior knowledge of human body structure using a coarse-to-fine approach, effectively extracting action features. Secondly, the Coordination Module utilizes contralateral and ipsilateral coordinated movements in human kinematics. Lastly, the Multi-scale Global Spatio-temporal Attention Module captures spatiotemporal features of different granularities and incorporates a causal convolution block and masked temporal attention to prevent non-causal relationships. This method achieved accuracy rates of 91.9% (Xsub), 96.3% (Xview), 88.5% (Xsub), and 90.3% (Xset) on NTU-RGB+D 60 and NTU-RGB+D 120, respectively.