ECCV-2024-Papers-Autonomous-Driving

We will promptly include more related works in this repository. Please stay tuned!!!

We also kindly invite you to our platform, Auto Driving Heart, for paper interpretation and sharing. If you would like to promote your work, please feel free to contact me.

1) End to End | 端到端自动驾驶

GenAD: Generative End-to-End Autonomous Driving

• paper: https://arxiv.org/pdf/2402.11502
• code: https://github.com/wzzheng/GenAD

CarFormer: Self-Driving with Learned Object-Centric Representations

• paper: https://arxiv.org/pdf/2407.15843
• code: https://github.com/Shamdan17/CarFormer

2）LLM Agent | 大语言模型智能体

DriveLM: Driving with Graph Visual Question Answering

• paper: https://arxiv.org/pdf/2312.14150
• code: https://github.com/OpenDriveLab/DriveLM

ELM: Embodied Understanding of Driving Scenarios

• paper: https://arxiv.org/pdf/2403.04593
• code: https://github.com/OpenDriveLab/ELM

Controllable Navigation Instruction Generation with Chain of Thought Prompting

• paper: coming soon
• code: https://github.com/refkxh/C-Instructor

Asynchronous Large Language Model Enhanced Planner for Autonomous Driving

• paper: coming soon
• code: coming soon

TOD3Cap: Towards 3D Dense Captioning in Outdoor Scenes

• paper: https://arxiv.org/pdf/2403.19589
• code: https://github.com/jxbbb/TOD3Cap

Global-Local Collaborative Inference with LLM for Lidar-Based Open-Vocabulary Detection

• paper: coming soon
• code: https://github.com/GradiusTwinbee/GLIS

Dolphins: Multimodal Language Model for Driving

• paper: https://arxiv.org/pdf/2312.00438
• code: https://github.com/SaFoLab-WISC/Dolphins

3）SSC: Semantic Scene Completion | 语义场景补全

Hierarchical Temporal Context Learning for Camera-based Semantic Scene Completion

• paper: https://arxiv.org/pdf/2407.02077
• code: https://github.com/Arlo0o/HTCL

Pyramid Diffusion for Fine 3D Large Scene Generation

• paper: https://arxiv.org/pdf/2311.12085
• code: https://github.com/yuhengliu02/pyramid-discrete-diffusion

4）OCC: Occupancy Prediction | 占用感知

Fully Sparse 3D Occupancy Prediction

• paper: https://arxiv.org/pdf/2312.17118
• code: https://github.com/MCG-NJU/SparseOcc

GaussianFormer: Scene as Gaussians for Vision-Based 3D Semantic Occupancy Prediction

• paper: https://arxiv.org/pdf/2405.17429
• code: https://github.com/huang-yh/GaussianFormer

Occupancy as Set of Points

• paper: https://arxiv.org/pdf/2407.04049
• code: https://github.com/hustvl/osp

Open Vocabulary 3D Scene Understanding via Geometry Guided Self-Distillation

• paper: https://arxiv.org/pdf/2407.13362
• code: https://github.com/Wang-pengfei/GGSD

ViewFormer: Exploring Spatiotemporal Modeling for Multi-View 3D Occupancy Perception via View-Guided Transformers

• paper: https://arxiv.org/pdf/2405.04299
• code: https://github.com/ViewFormerOcc/ViewFormer-Occ

nuCraft: Crafting High Resolution 3D Semantic Occupancy for Unified 3D Scene Understanding

• paper: https://www.ecva.net/papers/eccv_2024/papers_ECCV/papers/00730.pdf
• code: coming soon

5) World Model | 世界模型

OccWorld: 3D World Model for Autonomous Driving

• paper: https://arxiv.org/pdf/2311.16038
• code: https://github.com/wzzheng/OccWorld

Modelling Competitive Behaviors in Autonomous Driving Under Generative World Model

• paper: coming soon
• code: coming soon

DriveDreamer: Towards Real-world-driven World Models for Autonomous Driving

• paper: https://arxiv.org/pdf/2309.09777
• code: https://github.com/JeffWang987/DriveDreamer

6）HD-Mapping

MapTracker: Tracking with Strided Memory Fusion for Consistent Vector HD Mapping

• paper: https://arxiv.org/pdf/2403.15951
• code: https://github.com/woodfrog/maptracker

ADMap: Anti-disturbance framework for reconstructing online vectorized HD map

• paper: coming soon
• code: https://github.com/hht1996ok/ADMap

Accelerating Online Mapping and Behavior Prediction via Direct BEV Feature Attention

• paper: coming soon
• code: https://github.com/alfredgu001324/MapBEVPrediction

Leveraging Enhanced Queries of Point Sets for Vectorized Map Construction

• paper: https://arxiv.org/pdf/2402.17430
• code: https://github.com/HXMap/MapQR

7）Foundation Model

PartGLEE: A Foundation Model for Recognizing and Parsing Any Objects

• paper: coming soon
• code: coming soon

8）Robust Perception

Rethinking Data Augmentation for Robust LiDAR Semantic Segmentation in Adverse Weather

• paper: https://arxiv.org/pdf/2407.02286
• code: https://github.com/engineerJPark/LiDAR-DataAug4Weather

R2-Bench: Benchmarking the Robustness of Referring Perception Models under Perturbations

• paper: coming soon
• code: https://github.com/lxa9867/r2bench

9）3D Object Detection | 三维目标检测

Weakly Supervised 3D Object Detection via Multi-Level Visual Guidance

• paper: https://arxiv.org/pdf/2312.07530
• code: https://github.com/KuanchihHuang/VG-W3D

GraphBEV: Towards Robust BEV Feature Alignment for Multi-Modal 3D Object Detection

• paper: https://arxiv.org/pdf/2403.11848
• code: https://github.com/adept-thu/GraphBEV

RecurrentBEV: A Long-term Temporal Fusion Framework for Multi-view 3D Detection

• paper: coming soon
• code: https://github.com/lucifer443/RecurrentBEV

Ray Denoising: Depth-aware Hard Negative Sampling for Multi-view 3D Object Detection

• paper: https://arxiv.org/pdf/2402.03634
• code: https://github.com/LiewFeng/RayDN

MonoWAD: Weather-Adaptive Diffusion Model for Robust Monocular 3D Object Detection

• paper: coming soon
• code: https://github.com/VisualAIKHU/MonoWAD

DualBEV: CNN is All You Need in View Transformation

• paper: https://arxiv.org/pdf/2403.05402
• code: https://github.com/PeidongLi/DualBEV

OPEN: Object-wise Position Embedding for Multi-view 3D Object Detection

• paper: coming soon
• code: https://github.com/AlmoonYsl/OPEN

Make Your ViT-based Multi-view 3D Detectors Faster via Token Compression

• paper: coming soon
• code: coming soon

SEED: A Simple and Effective 3D DETR in Point Clouds

• paper: coming soon
• code: coming soon

Towards Stable 3D Object Detection

• paper: https://arxiv.org/pdf/2407.04305
• code: https://github.com/jbwang1997/StabilityIndex

FSD-BEV: Foreground Self-Distillation for Multi-view 3D Object Detection

• paper: coming soon
• code: https://github.com/CocoBoom/fsd-bev

HENet: Hybrid Encoding for End-to-end Multi-task 3D Perception from Multi-view Cameras

• paper: https://arxiv.org/pdf/2404.02517
• code: https://github.com/VDIGPKU/HENet

RepVF: A Unified Vector Fields Representation for Multi-task 3D Perception

• paper: https://arxiv.org/pdf/2407.10876
• code: https://github.com/jbji/RepVF

Approaching Outside: Scaling Unsupervised 3D Object Detection from 2D Scene

• paper: https://arxiv.org/abs/2407.08569
• code: https://github.com/Ruiyang-061X/LiSe

SimPB: A Single Model for 2D and 3D Object Detection from Multiple Cameras

• paper: https://arxiv.org/pdf/2403.10353
• code: https://github.com/nullmax-vision/SimPB

Interactive 3D Object Detection with Prompts

• paper: https://www.ecva.net/papers/eccv_2024/papers_ECCV/papers/02556.pdf
• code: coming soon

CSOT: Cross-Scan Object Transfer for Semi-Supervised LiDAR Object Detection

• paper: https://www.ecva.net/papers/eccv_2024/papers_ECCV/papers/02623.pdf
• code: https://github.com/JinglinZhan/CSOT

Approaching Outside: Scaling Unsupervised 3D Object Detection from 2D Scene

• paper: https://arxiv.org/pdf/2407.08569
• code: coming soon

10）Domain Adaptation & Test-Time Adaptation

Enhancing Source-Free Domain Adaptive Object Detection with Low-Confidence Pseudo-Label Distillation

• paper: coming soon
• code: https://github.com/junia3/LPLD

Fully Test-Time Adaptation for Monocular 3D Object Detection

• paper: coming soon
• code: https://github.com/Hongbin98/MonoTTA

Progressive Classifier and Feature Extractor Adaptation for Unsupervised Domain Adaptation on Point Clouds

• paper: https://arxiv.org/pdf/2303.01276
• code: https://github.com/xiaoyao3302/PCFEA

CMD: A Cross Mechanism Domain Adaptation Dataset for 3D Object Detection

• paper: coming soon
• code: https://github.com/im-djh/CMD

11）Cooperative Perception | 协同感知

Plug and Play: A Representation Enhanced Domain Adapter for Collaborative Perception

• paper: coming soon
• code: https://github.com/luotianyou349/PnPDA

Align before Collaborate: Mitigating Feature Misalignment for Robust Multi-Agent Perception

• paper: https://www.ecva.net/papers/eccv_2024/papers_ECCV/papers/00560.pdf
• code: coming soon

12）World Model

Neural Volumetric World Models for Autonomous Driving

• paper: https://www.ecva.net/papers/eccv_2024/papers_ECCV/papers/02571.pdf
• code: coming soon

13）Scene Flow Estimation | 场景流估计

4D Contrastive Superflows are Dense 3D Representation Learners

• paper: https://arxiv.org/pdf/2407.06190
• code: https://github.com/Xiangxu-0103/SuperFlow

SeFlow: A Self-Supervised Scene Flow Method in Autonomous Driving

• paper: https://arxiv.org/pdf/2407.01702
• code: https://github.com/KTH-RPL/SeFlow

I Can't Believe It's Not Scene Flow!

• paper: https://www.ecva.net/papers/eccv_2024/papers_ECCV/papers/02757.pdf
• code: https://github.com/kylevedder/BucketedSceneFlowEval

14）Point Cloud Semantic Segmentation| 点云

T-CorresNet: Template Guided 3D Point Cloud Completion with Correspondence Pooling Query Generation Strategy

• paper: coming soon
• code: https://github.com/df-boy/T-CorresNet

Dual-level Adaptive Self-Labeling for Novel Class Discovery in Point Cloud Segmentation

• paper: https://arxiv.org/pdf/2407.12489
• code: https://github.com/RikkiXu/NCD_PC

SFPNet: Sparse Focal Point Network for Semantic Segmentation on General LiDAR Point Clouds

• paper: https://arxiv.org/pdf/2407.11569
• code: coming soon

RAPiD-Seg: Range-Aware Pointwise Distance Distribution Networks for 3D LiDAR Segmentation

• paper: https://arxiv.org/pdf/2407.10159
• code: coming soon

ItTakesTwo: Leveraging Peer Representations for Semi-supervised LiDAR Semantic Segmentation

• paper: https://arxiv.org/pdf/2407.07171
• code: coming soon

Rethinking Data Augmentation for Robust LiDAR Semantic Segmentation in Adverse Weather

• paper: https://www.ecva.net/papers/eccv_2024/papers_ECCV/papers/02480.pdf
• code: https://github.com/engineerJPark/LiDARWeather

MUSES: The Multi-Sensor Semantic Perception Dataset for Driving under Uncertainty

• paper: https://arxiv.org/pdf/2401.12761
• code: https://github.com/timbroed/MUSES

Train Till You Drop: Towards Stable and Robust Source-free Unsupervised 3D Domain Adaptation

• paper: https://www.ecva.net/papers/eccv_2024/papers_ECCV/papers/02968.pdf
• code: https://github.com/valeoai/TTYD

T-MAE: Temporal Masked Autoencoders for Point Cloud Representation Learning

• paper: https://arxiv.org/pdf/2312.10217
• code: coming soon

15) Generative Model

RealGen: Retrieval Augmented Generation for Controllable Traffic Scenarios

• paper: https://arxiv.org/pdf/2312.13303
• code: https://realgen.github.io/

SLEDGE: Synthesizing Driving Environments with Generative Models and Rule-Based Traffic

• paper: https://arxiv.org/pdf/2403.17933
• code: https://github.com/autonomousvision/sledge

Photorealistic Object Insertion with Diffusion-Guided Inverse Rendering

• paper: https://arxiv.org/pdf/2408.09702
• code: https://research.nvidia.com/labs/toronto-ai/DiPIR/

16) Optical Flow

SEA-RAFT: Simple, Efficient, Accurate RAFT for Optical Flow

• paper: https://arxiv.org/pdf/2405.14793
• code: https://github.com/princeton-vl/SEA-RAFT

17）Radar | 毫米波雷达

Sparse Beats Dense: Rethinking Supervision in Radar-Camera Depth Completion

• paper: https://arxiv.org/pdf/2312.00844
• code: https://github.com/megvii-research/Sparse-Beats-Dense

18）Nerf Gaussian Splatting

Street Gaussians: Modeling Dynamic Urban Scenes with Gaussian Splatting

• paper: https://arxiv.org/pdf/2401.01339
• code: https://github.com/zju3dv/street_gaussians

MVSplat: Efficient 3D Gaussian Splatting from Sparse Multi-View Images

• paper: https://arxiv.org/pdf/2403.14627
• code: https://github.com/donydchen/mvsplat

GScream: Learning 3D Geometry and Feature Consistent Gaussian Splatting for Object Removal

• paper: https://arxiv.org/pdf/2404.13679
• code: https://github.com/W-Ted/GScream

BeNeRF: Neural Radiance Fields from a Single Blurry Image and Event Stream

• paper: coming soon
• code: https://github.com/WU-CVGL/BeNeRF

PreSight: Enhancing Autonomous Vehicle Perception with City-Scale NeRF Priors

• paper: https://arxiv.org/pdf/2403.09079
• code: https://github.com/yuantianyuan01/PreSight

GaussianImage: 1000 FPS Image Representation and Compression by 2D Gaussian Splatting

• paper: https://arxiv.org/pdf/2403.08551
• code: https://github.com/Xinjie-Q/GaussianImage

SG-NeRF: Neural Surface Reconstruction with Scene Graph Optimization

• paper: coming soon
• code: https://github.com/Iris-cyy/SG-NeRF

Disentangled Generation and Aggregation for Robust Radiance Fields

• paper: coming soon
• code: https://github.com/GaoHchen/Robust-Triplane

RPBG: Towards Robust Neural Point-based Graphics in the Wild

• paper: https://arxiv.org/pdf/2405.05663
• code: coming soon

19）Object Tracking | 目标跟踪

Beyond MOT: Semantic Multi-Object Tracking

• paper: coming soon
• code: https://github.com/HengLan/SMOT

3D Single-object Tracking in Point Clouds with High Temporal Variation

• paper: https://arxiv.org/pdf/2408.02049
• code: coming soon

OneTrack: Demystifying the Conflict Between Detection and Tracking in End-to-End 3D Trackers

• paper: https://www.ecva.net/papers/eccv_2024/papers_ECCV/papers/01174.pdf
• code: coming soon

Walker: Self-supervised Multiple Object Tracking by Walking on Temporal Object Appearance Graphs

• paper: https://www.ecva.net/papers/eccv_2024/papers_ECCV/papers/01205.pdf
• code: https://github.com/mattiasegu/walker

Boosting 3D Single Object Tracking with 2D Matching Distillation and 3D Pre-training

• paper: https://www.ecva.net/papers/eccv_2024/papers_ECCV/papers/01900.pdf
• code: coming soon

20）Lane Detection | 车道线检测

OMR: Occlusion-Aware Memory-Based Refinement for Video Lane Detection

• paper: https://arxiv.org/pdf/2408.07486
• code: https://github.com/dongkwonjin/OMR

RoadPainter: Points Are Ideal Navigators for Topology transformER

• paper: https://arxiv.org/pdf/2407.15349
• code: coming soon

21) Motion Prediction | 运动预测

22) Trajectory Prediction | 轨迹预测

VisionTrap: Vision-Augmented Trajectory Prediction Guided by Textual Descriptions

• paper: https://arxiv.org/pdf/2407.12345
• code: https://moonseokha.github.io/VisionTrap/

Risk-Aware Self-Consistent Imitation Learning for Trajectory Planning in Autonomous Driving

• paper: https://www.ecva.net/papers/eccv_2024/papers_ECCV/papers/02087.pdf
• code: coming soon

23) Depth Estimation | 深度估计

Leveraging Synthetic Data for Real-Domain High-Resolution Monocular Metric Depth Estimation

• paper: coming soon
• code: https://github.com/zhyever/PatchRefiner

ProDepth: Boosting Self-Supervised Multi-Frame Monocular Depth with Probabilistic Fusion

• paper: https://arxiv.org/pdf/2407.09303
• code: https://github.com/Sungmin-Woo/ProDepth

Diffusion Models for Monocular Depth Estimation: Overcoming Challenging Conditions

• paper: https://arxiv.org/pdf/2407.16698
• code: https://github.com/fabiotosi92/Diffusion4RobustDepth

Mono-ViFI: A Unified Learning Framework for Self-supervised Single- and Multi-frame Monocular Depth Estimation

• paper: https://arxiv.org/pdf/2407.14126
• code: https://github.com/LiuJF1226/Mono-ViFI

M2Depth: Self-supervised Two-Frame Multi-camera Metric Depth Estimation

• paper: https://arxiv.org/pdf/2405.02004
• code: https://heiheishuang.xyz/M2Depth/

24) Event Camera | 事件相机

25) Odometry

DVLO: Deep Visual-LiDAR Odometry with Local-to-Global Feature Fusion and Bi-Directional Structure Alignment

• paper: coming soon
• code: https://github.com/IRMVLab/DVLO

26）Normalized Object Coordinates

OmniNOCS: A unified NOCS dataset and model for 3D lifting of 2D objects

• paper: https://arxiv.org/pdf/2407.08711
• code: https://omninocs.github.io/

Postscript

This list of papers is primarily curated by Rujia Wang.

If you have any questions about the paper list, please do not hesitate to email me and [Auto Driving Heart Team] or open an issue on GitHub.