The field of computer vision and robotics is rapidly evolving, with a focus on developing more accurate and robust methods for object detection, tracking, and localization. Recent research has explored the use of multi-modal fusion, including camera-radar fusion, to improve the accuracy and reliability of these methods. Additionally, there has been a growing interest in developing datasets and benchmarks for specific applications, such as human-robot interaction and Martian digital elevation model prediction. Noteworthy papers in this area include C-DiffDet+, which introduces a novel approach to object detection by fusing global scene context with generative denoising, and InsFusion, which proposes a new method for instance-level LiDAR-camera fusion for 3D object detection. Other notable papers include Sem-RaDiff, which presents a diffusion-based 3D radar semantic perception framework, and CRAB, which introduces a camera-radar fusion-based 3D object detection and segmentation model.
Advancements in Computer Vision and Robotics
Sources
C-DiffDet+: Fusing Global Scene Context with Generative Denoising for High-Fidelity Object Detection
Towards Methane Detection Onboard Satellites
MVTrajecter: Multi-View Pedestrian Tracking with Trajectory Motion Cost and Trajectory Appearance Cost
Seeing through Unclear Glass: Occlusion Removal with One Shot
PrediTree: A Multi-Temporal Sub-meter Dataset of Multi-Spectral Imagery Aligned With Canopy Height Maps
Sem-RaDiff: Diffusion-Based 3D Radar Semantic Perception in Cluttered Agricultural Environments
Comparative Evaluation of Traditional and Deep Learning Feature Matching Algorithms using Chandrayaan-2 Lunar Data
PropVG: End-to-End Proposal-Driven Visual Grounding with Multi-Granularity Discrimination
Towards an Accurate and Effective Robot Vision (The Problem of Topological Localization for Mobile Robots)
MonoGlass3D: Monocular 3D Glass Detection with Plane Regression and Adaptive Feature Fusion
CRAB: Camera-Radar Fusion for Reducing Depth Ambiguity in Backward Projection based View Transformation
S-LAM3D: Segmentation-Guided Monocular 3D Object Detection via Feature Space Fusion
DCReg: Decoupled Characterization for Efficient Degenerate LiDAR Registration
Cross3DReg: Towards a Large-scale Real-world Cross-source Point Cloud Registration Benchmark
LiHRA: A LiDAR-Based HRI Dataset for Automated Risk Monitoring Methods
MCTED: A Machine-Learning-Ready Dataset for Digital Elevation Model Generation From Mars Imagery
Dual-Thresholding Heatmaps to Cluster Proposals for Weakly Supervised Object Detection
InsFusion: Rethink Instance-level LiDAR-Camera Fusion for 3D Object Detection
Sparse BEV Fusion with Self-View Consistency for Multi-View Detection and Tracking
CLAP: Clustering to Localize Across n Possibilities, A Simple, Robust Geometric Approach in the Presence of Symmetries