CVPR 2026 论文和开源项目合集(Papers with Code)

CVPR 2026 decisions are now available on OpenReview！25.42% = 4090 / 16092

注1：欢迎各位大佬提交issue，分享CVPR 2026论文和开源项目！

注2：关于往年CV顶会论文以及其他优质CV论文和大盘点，详见： https://github.com/amusi/daily-paper-computer-vision

• ICCV 2025
• ECCV 2024

欢迎扫码加入【CVer学术交流群】，可以获取CVPR 2026等最前沿工作！这是最大的计算机视觉AI知识星球！每日更新，第一时间分享最新最前沿的计算机视觉、AIGC、扩散模型、多模态、深度学习、自动驾驶、医疗影像和遥感等方向的学习资料，快加入学起来！

【CVPR 2026 论文开源目录】

• 3DGS(Gaussian Splatting)
• Agent)
• Avatars
• Backbone
• CLIP
• Mamba
• Embodied AI
• GAN
• GNN
• 多模态大语言模型(MLLM)
• 大语言模型(LLM)
• 具身智能(Embodied AI)
• 空间智能(Spatial Intelligence
• NAS
• OCR
• NeRF
• DETR
• 扩散模型(Diffusion Models)
• ReID(重识别)
• 长尾分布(Long-Tail)
• Vision Transformer
• 视觉和语言(Vision-Language)
• 自监督学习(Self-supervised Learning)
• 数据增强(Data Augmentation)
• 目标检测(Object Detection)
• 异常检测(Anomaly Detection)
• 目标跟踪(Visual Tracking)
• 语义分割(Semantic Segmentation)
• 实例分割(Instance Segmentation)
• 全景分割(Panoptic Segmentation)
• 医学图像(Medical Image)
• 医学图像分割(Medical Image Segmentation)
• 视频目标分割(Video Object Segmentation)
• 视频实例分割(Video Instance Segmentation)
• 参考图像分割(Referring Image Segmentation)
• 图像抠图(Image Matting)
• 图像编辑(Image Editing)
• Low-level Vision
• 超分辨率(Super-Resolution)
• 去噪(Denoising)
• 去模糊(Deblur)
• 自动驾驶(Autonomous Driving)
• 3D点云(3D Point Cloud)
• 3D目标检测(3D Object Detection)
• 3D语义分割(3D Semantic Segmentation)
• 3D目标跟踪(3D Object Tracking)
• 3D语义场景补全(3D Semantic Scene Completion)
• 3D配准(3D Registration)
• 3D人体姿态估计(3D Human Pose Estimation)
• 3D人体Mesh估计(3D Human Mesh Estimation)
• 3D Visual Grounding(3D视觉定位)
• 医学图像(Medical Image)
• 图像生成(Image Generation)
• 视频生成(Video Generation)
• 3D生成(3D Generation)
• 视频理解(Video Understanding)
• 行为检测(Action Detection)
• 遥感(Remote)
• 文本检测(Text Detection)
• 知识蒸馏(Knowledge Distillation)
• 模型剪枝(Model Pruning)
• 图像压缩(Image Compression)
• 视频压缩(Video Compression)
• 三维重建(3D Reconstruction)
• 深度估计(Depth Estimation)
• 轨迹预测(Trajectory Prediction)
• 车道线检测(Lane Detection)
• 图像描述(Image Captioning)
• 视觉问答(Visual Question Answering)
• 手语识别(Sign Language Recognition)
• 视频预测(Video Prediction)
• 新视点合成(Novel View Synthesis)
• Zero-Shot Learning(零样本学习)
• 立体匹配(Stereo Matching)
• 特征匹配(Feature Matching)
• 暗光图像增强(Low-light Image Enhancement)
• 场景图生成(Scene Graph Generation)
• 图像检索(Image Retrieval)
• 风格迁移(Style Transfer)
• 隐式神经表示(Implicit Neural Representations)
• 图像质量评价(Image Quality Assessment)
• 视频质量评价(Video Quality Assessment)
• 压缩感知(Compressive Sensing)
• 数据集(Datasets)
• 新任务(New Tasks)
• 其他(Others)

3DGS(Gaussian Splatting)

Dropping Anchor and Spherical Harmonics for Sparse-view Gaussian Splatting

• Paper: https://arxiv.org/abs/2602.20933
• Code:
• Project: https://sk-fun.fun/DropAnSH-GS

Topology-Aware Gaussian Splatting for Dynamic Mesh Modeling and Tracking

• Paper: https://arxiv.org/abs/2512.01329
• Project: https://haza628.github.io/tagSplat/

FastGS: Training 3D Gaussian Splatting in 100 Seconds

• Paper: https://arxiv.org/pdf/2511.04283
• Code: https://github.com/fastgs/FastGS
• Project: https://fastgs.github.io/

Agent

Avatars

Backbone

CLIP

Mamba

GAN

OCR

NeRF

DETR

Prompt

多模态大语言模型(MLLM)

Circuit Tracing in Vision-Language Models: Understanding the Internal Mechanisms of Multimodal Thinking

• Paper: https://arxiv.org/abs/2602.20330
• Code: https://github.com/UIUC-MONET/vlm-circuit-tracing

UniM: A Unified Any-to-Any Interleaved Multimodal Benchmark

• Paper: https://arxiv.org/abs/2603.05075
• Code:
• Project: https://any2any-mllm.github.io/unim/

大语言模型(LLM)

具身智能(Embodied AI)

Wanderland: Geometrically Grounded Simulation for Open-World Embodied AI

• Paper: https://arxiv.org/abs/2511.20620
• Code: https://github.com/ai4ce/wanderland
• Project: https://ai4ce.github.io/wanderland/

空间智能(Spatial Intelligence)

Spatial-SSRL: Enhancing Spatial Understanding via Self-Supervised Reinforcement Learning

• Paper: https://arxiv.org/abs/2510.27606
• Code: https://github.com/InternLM/Spatial-SSRL
• Model: https://huggingface.co/internlm/Spatial-SSRL-7B

NAS

ReID(重识别)

MOS: Mitigating Optical-SAR Modality Gap for Cross-Modal Ship Re-Identification

• Paper: https://arxiv.org/abs/2512.03404
• Code: https://github.com/yjzhao1019/MOS

扩散模型(Diffusion Models)

Vision Transformer

视觉和语言(Vision-Language)

StructXLIP: Enhancing Vision-language Models with Multimodal Structural Cues

• Paper: https://arxiv.org/abs/2602.20089
• Code: https://github.com/intelligolabs/StructXLIP

ApET: Approximation-Error Guided Token Compression for Efficient VLMs

• Paper: https://arxiv.org/abs/2602.19870
• Code: https://github.com/MaQianKun0/ApET

Circuit Tracing in Vision-Language Models: Understanding the Internal Mechanisms of Multimodal Thinking

• Paper: https://arxiv.org/abs/2602.20330
• Code: https://github.com/UIUC-MONET/vlm-circuit-tracing

目标检测(Object Detection)

异常检测(Anomaly Detection)

目标跟踪(Object Tracking)

医学图像(Medical Image)

医学图像分割(Medical Image Segmentation)

MedCLIPSeg: Probabilistic Vision–Language Adaptation for Data-Efficient and Generalizable Medical Image Segmentation

• Paper: https://arxiv.org/abs/2602.20423
• Code: https://github.com/HealthX-Lab/MedCLIPSeg
• Project: https://tahakoleilat.github.io/MedCLIPSeg

自动驾驶(Autonomous Driving)

Open-Vocabulary Domain Generalization in Urban-Scene Segmentation

• Paper: https://arxiv.org/pdf/2602.18853
• Code: https://github.com/DZhaoXd/s2_corr

U4D: Uncertainty-Aware 4D World Modeling from LiDAR Sequences

• Paper: https://arxiv.org/abs/2512.02982
• Code: https://github.com/worldbench/U4D

3D点云(3D-Point-Cloud)

CLIPoint3D: Language-Grounded Few-Shot Unsupervised 3D Point Cloud Domain Adaptation

• Paper: https://arxiv.org/abs/2602.20409
• Code: https://github.com/SarthakM320/CLIPoint3D

3D目标检测(3D Object Detection)

3D语义分割(3D Semantic Segmentation)

Low-level Vision

超分辨率(Super-Resolution)

去噪(Denoising)

图像去噪(Image Denoising)

3D人体姿态估计(3D Human Pose Estimation)

#3D Visual Grounding(3D视觉定位)

图像生成(Image Generation)

ExpPortrait: Expressive Portrait Generation via Personalized Representation

• Paper: https://arxiv.org/abs/2602.19900
• Code:

视频生成(Video Generation)

图像编辑(Image Editing)

视频编辑(Video Editing)

3D生成(3D Generation)

3D重建(3D Reconstruction)

tttLRM: Test-Time Training for Long Context and Autoregressive 3D Reconstruction

• Project: https://cwchenwang.github.io/tttLRM/
• Paper: https://arxiv.org/abs/2602.20160
• Code: https://github.com/cwchenwang/tttLRM

Flow3r: Factored Flow Prediction for Scalable Visual Geometry Learning

• Project: https://flow3r-project.github.io/
• Paper: https://arxiv.org/abs/2602.20157
• Code: https://github.com/Kidrauh/flow3r

RAP: Fast Feedforward Rendering-Free Attribute-Guided Primitive Importance Score Prediction for Efficient 3D Gaussian Splatting Processing

• Paper: https://arxiv.org/abs/2602.19753
• Code: https://github.com/yyyykf/RAP

人体运动生成(Human Motion Generation)

视频理解(Video Understanding)

遥感(Remote)

Brewing Stronger Features: Dual-Teacher Distillation for Multispectral Earth Observation

• Paper: https://arxiv.org/abs/2602.19863
• Code: None

知识蒸馏(Knowledge Distillation)

深度估计(Depth Estimation)

立体匹配(Stereo Matching)

暗光图像增强(Low-light Image Enhancement)

图像压缩(Image Compression)](#IC)

视频压缩(Video Compression)](#VC)

UniComp: Rethinking Video Compression Through Informational Uniqueness

• Paper: https://arxiv.org/abs/2512.03575
• Code: https://github.com/TimeMarker-LLM/UniComp

场景图生成(Scene Graph Generation)

图像检索(Image Retrieval)

**PinPoint: Evaluation of Composed Image Retrieval with Explicit Negatives, Multi-Image Queries, and Paraphrase Testing **

• Paper: https://arxiv.org/abs/2603.04598
• Code:

风格迁移(Style Transfer)

图像质量评价(Image Quality Assessment)

视频质量评价(Video Quality Assessment)

压缩感知(Compressive Sensing)

数据集(Datasets)

其他(Others)

Decoupling Defense Strategies for Robust Image Watermarking

• Paper: https://arxiv.org/abs/2602.20053
• Code: None

Multi-Modal Representation Learning via Semi-Supervised Rate Reduction for Generalized Category Discovery

• Paper: https://arxiv.org/abs/2602.19910
• Code:

The Invisible Gorilla Effect in Out-of-distribution Detection

• Paper: https://arxiv.org/abs/2602.20068
• Code: https://github.com/HarryAnthony/Invisible_Gorilla_Effect

SimLBR: Learning to Detect Fake Images by Learning to Detect Real Images

• Paper: https://arxiv.org/abs/2602.20412
• Code:

RecoverMark: Robust Watermarking for Localization and Recovery of Manipulated Faces

• Paper: https://arxiv.org/abs/2602.20618
• Code:

Probing and Bridging Geometry-Interaction Cues for Affordance Reasoning in Vision Foundation Models

• Paper:
• Code:

GEM-TFL: Bridging Weak and Full Supervision for Forgery Localization through EM-Guided Decomposition and Temporal Refinement

• Paper: https://arxiv.org/abs/2603.05095
• Code:

FOZO: Forward-Only Zeroth-Order Prompt Optimization for Test-Time Adaptation

• Paper: https://arxiv.org/abs/2603.04733
• Code: https://github.com/eVI-group-SCU/FOZO

**Mitigating Instance Entanglement in Instance-Dependent Partial Label Learning **

• Paper: https://arxiv.org/abs/2603.04825
• Code: https://github.com/RyanZhaoIc/CAD