The field of computer vision is rapidly advancing, with a strong focus on developing innovative solutions for autonomous systems. Recent developments have seen significant improvements in areas such as 3D object detection, monocular depth estimation, and image fusion. Researchers have proposed novel architectures and techniques, including the use of quaternion neural networks, adaptive statistical independence testing, and chain-of-prediction models, to tackle complex challenges in computer vision. These advancements have the potential to enhance the performance and efficiency of autonomous vehicles, robots, and other systems that rely on computer vision. Noteworthy papers in this area include the Efficient On-Chip Implementation of 4D Radar-Based 3D Object Detection, which achieved real-time processing on a low-power embedded environment, and the MonoCoP approach, which leveraged a chain-of-prediction to predict 3D attributes sequentially and conditionally. Overall, the field of computer vision is making rapid progress, driven by the development of new techniques and architectures that can efficiently and accurately process complex visual data.
Advances in Computer Vision for Autonomous Systems
Sources
Efficient On-Chip Implementation of 4D Radar-Based 3D Object Detection on Hailo-8L
LMDepth: Lightweight Mamba-based Monocular Depth Estimation for Real-World Deployment
Efficient Vision-based Vehicle Speed Estimation
Edge Detection based on Channel Attention and Inter-region Independence Test
Edge-preserving Image Denoising via Multi-scale Adaptive Statistical Independence Testing
CMAWRNet: Multiple Adverse Weather Removal via a Unified Quaternion Neural Architecture
DriveNetBench: An Affordable and Configurable Single-Camera Benchmarking System for Autonomous Driving Networks
Quaternion Multi-focus Color Image Fusion
Quaternion Infrared Visible Image Fusion
RGBX-DiffusionDet: A Framework for Multi-Modal RGB-X Object Detection Using DiffusionDet
An Enhanced YOLOv8 Model for Real-Time and Accurate Pothole Detection and Measurement
RGB-Event Fusion with Self-Attention for Collision Prediction
Predicting Road Surface Anomalies by Visual Tracking of a Preceding Vehicle
DFVO: Learning Darkness-free Visible and Infrared Image Disentanglement and Fusion All at Once
MonoCoP: Chain-of-Prediction for Monocular 3D Object Detection
Lightweight RGB-D Salient Object Detection from a Speed-Accuracy Tradeoff Perspective
Adaptive Contextual Embedding for Robust Far-View Borehole Detection