Kitti
PulseAugur coverage of Kitti — every cluster mentioning Kitti across labs, papers, and developer communities, ranked by signal.
11 day(s) with sentiment data
-
MetricTrail system enhances 3D reconstruction from long videos
Researchers have introduced MetricTrail, a novel system designed for accurate camera motion estimation and dense 3D reconstruction from extended monocular video sequences. The system enhances trajectory accuracy by inco…
-
New method detects pedestrian occlusion using monocular distance cues
Researchers have developed a novel method for detecting pedestrian occlusion in monocular vision systems, which is crucial for autonomous robots. This technique leverages the disagreement between two existing distance c…
-
New CLOCs-R framework fuses radar, camera, and LiDAR for long-range object detection
Researchers have developed CLOCs-R, a new framework that enhances long-range 3D object detection for autonomous vehicles by integrating radar data with camera and LiDAR information. The system accumulates radar returns …
-
Wavelet Phase Diffusion enhances sim-to-real translation with localized phase injection
Researchers have developed Wavelet Phase Diffusion (WPD), a novel method for translating synthetic images and videos into photorealistic real-world appearances. Unlike previous approaches that struggle with computationa…
-
New C-DFF method enhances autonomous driving validation in simulations
Researchers have developed a new method called Concept-Decisive Feature Fidelity (C-DFF) to improve the validation of autonomous driving systems. C-DFF addresses the gap where real and synthetic images might produce sim…
-
New method evaluates ML LiDAR safety targets across conditions
A new paper proposes an uncertainty evaluation method to support the Safety Of The Intended Functionality (SOTIF) analysis for machine learning object detectors used in LiDAR systems. The method aims to determine if per…
-
New STATIC model enhances video depth estimation with frame-to-frame consistency
Researchers have developed STATIC, a novel model for video depth estimation that improves temporal consistency across frames. Unlike previous methods that rely on external motion priors, STATIC independently learns cons…
-
LiDAR perception for autonomous driving made robust and efficient
Researchers have developed a new method for provisioning robust and efficient LiDAR perception systems for autonomous driving. This approach addresses both point-cloud corruption and resource constraints by training det…
-
New method enhances LiDAR perception for autonomous driving
Researchers have developed a new method for provisioning robust and efficient LiDAR perception systems for autonomous driving. This approach addresses both the challenge of maintaining accuracy under point-cloud corrupt…
-
MonoMIP enhances monocular 3D object detection with BEV features
Researchers have developed MonoMIP, a novel framework for monocular 3D object detection that enhances accuracy by incorporating Bird's-Eye View (BEV) features. This method utilizes Multi-Height Inverse Perspective Mappi…
-
New Bayesian framework enhances online 3D reconstruction for long video sequences
Researchers have developed BAT3R, a novel framework designed to improve online 3D reconstruction for long video sequences. This framework addresses the progressive degradation seen in existing recurrent architectures by…
-
GeoBridge framework evolves transferable decoder topologies for geometric vision
Researchers have introduced GeoBridge, an evolutionary framework designed to discover transferable decoder topologies for geometric vision tasks. This system optimizes internal computations within decoders, allowing the…
-
New AI models advance 3D world reconstruction and modeling
Researchers have developed new methods for 3D world modeling and reconstruction from images. GeoState focuses on creating view-decoupled geometry states for predictive modeling, separating scene dynamics from camera var…
-
JMT-YOLO framework enhances object detection across domain shifts
Researchers have introduced JMT-YOLO, a new framework designed to improve domain adaptive object detection (DAOD). This method addresses performance degradation caused by distribution shifts between training and testing…
-
MemReg network improves LiDAR registration using temporal data
Researchers have developed MemReg, a novel network for streaming outdoor LiDAR point cloud registration that utilizes hybrid memory buffers to capture temporal correlations. This approach leverages long-term temporal da…
-
Drive3R model advances long-horizon 3D reconstruction for autonomous driving
Researchers have introduced Drive3R, a novel geometric foundation model designed for autonomous driving that addresses the computational challenges of long-horizon 3D reconstruction. The model utilizes block-wise causal…
-
DVD method enhances MLLM perception with efficient vector decoding
Researchers have introduced Dynamic Vector Decoding (DVD), a novel method designed to enhance the efficiency of multimodal large language models (MLLMs) in perception tasks. DVD addresses limitations in current approach…
-
New MAGIC framework improves unsupervised stereo matching with visibility asymmetry
Researchers have developed MAGIC, a novel multi-baseline geometric consistency framework designed to improve unsupervised stereo matching, particularly in occluded regions. This method uses a teacher-student model with …
-
LePoKet framework enhances robotic vision with learnable knowledge transfer
Researchers have developed LePoKet, a novel framework for knowledge transfer in robotic vision systems. This method optimizes interaction parameters within a block-wise interface, enabling learnable parameter optimizati…
-
SAVTrack improves 3D LiDAR tracking with selective vote aggregation · 2 sources tracked
Researchers have developed SAVTrack, a new framework for 3D single object tracking in LiDAR point clouds that addresses challenges posed by sparse and incomplete data. The system employs Selective Vote Aggregation (SAV)…