The recent advancements in object detection and segmentation have significantly pushed the boundaries of what is possible in both closed- and open-set scenarios. A notable trend is the integration of language prompts to enhance the generalization and performance of models, particularly in open-vocabulary and unknown object detection tasks. Researchers are focusing on developing efficient prompt-guided mechanisms that leverage both textual and visual information to improve alignment and reduce bias. Additionally, there is a growing emphasis on lightweight and real-time frameworks that can be deployed in resource-constrained environments, such as robotics. These frameworks aim to decouple feature alignment and streamline computational processes, making them more suitable for practical applications. Interactive segmentation methods are also evolving, with new approaches that utilize sequence information and in-context guidance to reduce user interaction and improve segmentation accuracy. Overall, the field is moving towards more versatile, efficient, and user-friendly solutions that can handle a broader range of tasks and environments.