The field of underwater scene understanding and maritime object tracking is rapidly advancing with the development of new models and techniques. Researchers are focusing on improving the accuracy and robustness of underwater scene understanding models, particularly in visually-degraded environments. The use of multimodal models, such as those that combine sonar and visual data, is becoming increasingly popular. Additionally, there is a growing interest in developing more efficient and effective methods for maritime multi-object tracking, including the use of parallel tracking architectures and reversible columnar detection networks. Noteworthy papers in this area include NAUTILUS, which introduces a large multimodal model for underwater scene understanding, and DMSORT, which proposes an efficient parallel maritime multi-object tracking architecture. SonarSweep is also notable for its novel approach to fusing sonar and vision for robust 3D reconstruction. These advancements have the potential to significantly improve the performance of underwater and maritime applications, such as autonomous underwater vehicles and maritime surveillance systems.