The field of autonomous systems and computer vision is rapidly evolving, with a focus on developing more accurate and efficient methods for perception, prediction, and decision-making. Recent research has emphasized the importance of incorporating uncertainty and temporal information into models, as well as leveraging large-scale datasets and semi-supervised learning techniques to improve performance. Notably, innovative approaches to trajectory planning, driver behavior classification, and object detection have been proposed, demonstrating significant improvements over existing methods.
Some noteworthy papers in this area include: Occupancy-aware Trajectory Planning for Autonomous Valet Parking, which presents a novel approach to predicting future parking spot occupancy and integrated planning. Classification of Driver Behaviour Using External Observation Techniques for Autonomous Vehicles, which introduces a vision-based system for detecting indicators of distraction and impairment. A Co-Training Semi-Supervised Framework Using Faster R-CNN and YOLO Networks for Object Detection, which combines Faster R-CNN and YOLO networks for precise localization and global context in densely packed retail environments. Weakly and Self-Supervised Class-Agnostic Motion Prediction for Autonomous Driving, which investigates weakly and self-supervised class-agnostic motion prediction from LiDAR point clouds. Advancing Real-World Parking Slot Detection with Large-Scale Dataset and Semi-Supervised Baseline, which constructs a large-scale parking slot detection dataset and develops a semi-supervised baseline for parking slot detection. Road Obstacle Video Segmentation, which demonstrates that road-obstacle segmentation is inherently temporal and proposes strong baseline methods based on vision foundation models. Pseudo-Label Enhanced Cascaded Framework, which presents a solution for Complex Video Object Segmentation based on the SAM2 framework and pseudo-labeling strategy. PRISM: Product Retrieval In Shopping Carts using Hybrid Matching, which proposes a hybrid method for product retrieval in retail settings by leveraging the advantages of both vision-language model-based and pixel-wise matching approaches.
