CVPR2025

CVPR论文列表

大论文相关，abstact

• SphereUFormer: A U-Shaped Transformer for Spherical 360 Perception
  对360rgb图的深度进行估计

• CroCoDL: Cross-device Collaborative Dataset for Localization(没有）

• SemAlign3D: Semantic Correspondence between RGB-Images through Aligning 3D Object-Class Representations
  用点云的形式对应出两张图上语义相同的物体

• PlanarSplatting: Accurate Planar Surface Reconstruction in 3 Minutes
  3维重建并且新视角合成

• A General Adaptive Dual-level Weighting Mechanism for Remote Sensing Pansharpening
  遥感图像锐化（跟超分的区别：超分input为一张图，该task input为两张图，一张带颜色的低分辨率多光谱图，一张不带颜色的高分辨率纹理图

• Towards Generalizable Scene Change Detection
  没搞懂这个的意义在于？原文： The ability to accurately identify meaningful changes in a scene across different time steps—despite challenges such as illumination variations, seasonal changes, and weather conditions—plays a key role in the system’s effectiveness and reliability.

• Towards Long-Horizon Vision-Language Navigation: Platform, Benchmark and Method
  这个用language来提出task，思维链拆分成序列的任务，然后利用vision information去构建。无人机等导航任务可以用

• T-FAKE: Synthesizing Thermal Images for Facial Landmarking
  人脸标定

• Opportunistic Single-Photon Time of Flight（没有）

• AerialMegaDepth: Learning Aerial-Ground Reconstruction and View Synthesis
  空地视角重建与新视角合成

• AG-VPReID: A Challenging Large-Scale Benchmark for Aerial-Ground Video-based Person Re-Identification
  空地视角人re-identification，所以可以用作空地视角的物体-reidentification

• Object-aware Sound Source Localization via Audio-Visual Scene Understanding（没有）

• light3R-SfM: Towards Feed-forward Structure-from-Motion
  无序的一系列图片输入，3维点云图输出

• High Temporal Consistency through Semantic Similarity Propagation in Semi-Supervised Video Semantic Segmentation for Autonomous Flight

• EvOcc: Accurate Semantic Occupancy for Automated Driving Using Evidence Theory（没有）

• RUBIK: A Structured Benchmark for Image Matching across Geometric Challenges
  图像匹配–图像重复区域、放缩比例、视角变换角度

• SplatFlow: Self-Supervised Dynamic Gaussian Splatting in Neural Motion Flow Field for Autonomous Driving

• [x]Satellite to GroundScape - Large-scale Consistent Ground View Generation from Satellite Views
  一个目标物体从卫星视角到地面视角到转换，构建地面视角

• **XLRS-Bench: Could Your Multimodal LLMs Understand Extremely Large Ultra-High-Resolution Remote Sensing Imagery? **
  大模型对超高分辨率的遥感突袭那个的理解

• Satellite Observations Guided Diffusion Model for Accurate Meteorological States at Arbitrary Resolution
  气象预测 ，用的是卫星图不是rgb或者多光谱

• ABBSPO: Adaptive Bounding Box Scaling and Symmetric Prior based Orientation Prediction for Detecting Aerial Image Objects （没有）

• Multi-Modal Aerial-Ground Cross-View Place Recognition with Neural ODEs（没有）

• RobSense: A Robust Multi-modal Foundation Model for Remote Sensing with Static, Temporal, and Incomplete Data Adaptability （没有）

• SeCap: Self-Calibrating and Adaptive Prompts for Cross-view Person Re-Identification in Aerial-Ground Networks
  Person Re-Identification in Aerial-Ground Networks

• ROS-SAM: High-Quality Interactive Segmentation for Remote Sensing Moving Object
  还是目标检测，但是主要针对的类别是动目标（飞机、车、轮船），难点是目标小

• Detection-Friendly Nonuniformity Correction: A Union Framework for Infrared UAV Target Detection
  红外无人机检测，提供了一个新数据集和检测框架

• JarvisIR: Elevating Autonomous Driving Perception with Intelligent Image Restoration
  面向自动驾驶感知在不同天气情况下的图像退化问题，提出一个基于VLM的系统用于合理选择现有的专家修复模型。引入新数据集和人类反馈。

• The Change You Want To Detect: Semantic Change Detection In Earth Observation With Hybrid Data Generationf

• Effective Cloud Removal for Remote Sensing Images by an Improved Mean-Reverting Denoising Model with Elucidated Design Space

• Towards Satellite Image Road Graph Extraction: A Global-Scale Dataset and A Novel Method

• Learning Occlusion-Robust Vision Transformers for Real-Time UAV Tracking

• UCM-VeID V2: A Richer Dataset and A Pre-training Method for UAV Cross-Modality Vehicle Re-Identification

• Galaxy Walker: Geometry-aware VLMs For Galaxy-scale Understanding

• Exact: Exploring Space-Time Perceptive Clues for Weakly Supervised Satellite Image Time Series Semantic Segmentation

• Horizon-GS: Unified 3D Gaussian Splatting for Large-Scale Aerial-to-Ground Scenes

low-leve相关

• Uncertainty-guided Perturbation for Image Super-Resolution Diffusion Model
• Reconstructing Animals and the Wild
• Training Data Provenance Verification: Did Your Model Use Synthetic Data from My Generative Model for Training?
• Latent Space Super-Resolution for Higher-Resolution Image Generation with Diffusion Models
• Pathways on the Image Manifold: Image Editing via Video Generation
• Augmenting Perceptual Super-Resolution via Image Quality Predictors
• MaDCoW: Marginal Distortion Correction for Wide-Angle Photography with Arbitrary Objects
• Localized Concept Erasure for Text-to-Image Diffusion Models Using Training-Free Gated Low-Rank Adaptation
• UniRestore: Unified Perceptual and Task-Oriented Image Restoration Model Using Diffusion Prior
• LSNet: See Large, Focus Small
• Evaluating Model Perception of Color Illusions in Photorealistic Scenes
• Improving Diffusion Inverse Problem Solving with Decoupled Noise Annealing
• Degradation-Aware Feature Perturbation for All-in-One Image Restoration
• Good, Cheap, and Fast: Overfitted Image Compression with Wasserstein Distortion
• Edge-SD-SR: Low Latency and Parameter Efficient On-device Super-Resolution with Stable Diffusion via Bidirectional Conditioning
• A Regularization-Guided Equivariant Approach for Image Restoration
• The Power of Context: How Multimodality Improves Image Super-Resolution
• GenDeg: Diffusion-based Degradation Synthesis for Generalizable All-In-One Image Restoration
• Exposure-slot: Exposure-centric Representations Learning with Slot-in-Slot Attention for Region-aware Exposure Correction
• Positive2Negative: Breaking the Information-Lossy Barrier in Self-Supervised Single Image Denoising
• MaIR: A Locality- and Continuity-Preserving Mamba for Image Restoration
• Decouple Distortion from Perception: Region Adaptive Diffusion for Extreme-low Bitrate Perception Image Compression
• Rotation-Equivariant Self-Supervised Method in Image Denoising
• CATANet: Efficient Content-Aware Token Aggregation for Lightweight Image Super-Resolution
• Towards RAW Object Detection in Diverse Conditions
• ACL: Activating Capability of Linear Attention for Image Restoration
• Image Quality Assessment: Investigating Causal Perceptual Effects with Abductive Counterfactual Inference
• Progressive Focused Transformer for Single Image Super-Resolution
• HIIF: Hierarchical Encoding based Implicit Image Function for Continuous Super-resolution
• Efficient Diffusion as Low Light Enhance
• ADD: Attribution-Driven Data Augmentation Framework for Boosting Image Super-Resolution
• Bridging the Gap between Gaussian Diffusion Models and Universal Quantization for Image Compression
• Tokenize Image Patches: Global Context Fusion for Effective Haze Removal in Large Images
• Adapting Text-to-Image Generation with Feature Difference Instruction for Generic Image Restoration
• TSD-SR: One-Step Diffusion with Target Score Distillation for Real-World Image Super-Resolution
• Do Computer Vision Foundation Models Learn the Low-level Characteristics of the Human Visual System?
• Perceptual Video Compression with Neural Wrapping
• From Zero to Detail: Deconstructing Ultra-High-Definition Image Restoration from Progressive Spectral Perspective
• High Dynamic Range Video Compression: A Large-Scale Benchmark Dataset and A Learned Bit-depth Scalable Compression Algorithm
• DL2G: Degradation-guided Local-to-Global Restoration for Eyeglass Reflection Removal
• FLAVC: Learned Video Compression with Feature Level Attention
• Binarized Mamba-Transformer for Lightweight Quad Bayer HybridEVS Demosaicing
• ABC-Former: Auxiliary Bimodal Cross-domain Transformer with Interactive Channel Attention for White Balance
• Reversing Flow for Image Restoration
• Adaptive Rectangular Convolution for Remote Sensing Pansharpening
• DnLUT: Ultra-Efficient Color Image Denoising via Channel-Aware Lookup Tables
• PassionSR: Post-Training Quantization with Adaptive Scale in One-Step Diffusion based Image Super-Resolution
• Proximal Algorithm Unrolling: Flexible and Efficient Reconstruction Networks for Single-Pixel Imaging
• GCC: Generative Color Constancy via Diffusing a Color Checker
• WISE: A Framework for Gigapixel Whole-Slide-Image Lossless Compression
• Visual-Instructed Degradation Diffusion for All-in-One Image Restoration
• Can Text-to-Video Generation help Video-Language Alignment?
• VoCo-LLaMA: Towards Vision Compression with Large Language Models
• URWKV: Unified RWKV Model with Multi-state Perspective for Low-light Image Restoration
• MambaIC: State Space Models for High-Performance Learned Image Compression
• HVI: A New Color Space for Low-light Image Enhancement
• OSMamba: Omnidirectional Spectral Mamba with Dual-Domain Prior Generator for Exposure Correction
• Adversarial Diffusion Compression for Real-World Image Super-Resolution
• Diffusion-based Event Generation for High-Quality Image Deblurring
• CoA: Towards Real Image Dehazing via Compression-and-Adaptation
• PICD: Versatile Perceptual Image Compression with Diffusion Rendering
• UHD-processer: Unified UHD Image Restoration with Progressive Frequency Learning and Degradation-aware Prompts
• QMambaBSR: Burst Image Super-Resolution with Query State Space Model
• Balanced Rate-Distortion Optimization in Learned Image Compression
• UltraFusion: Ultra High Dynamic Imaging using Exposure Fusion
• Frequency-Biased Synergistic Design for Image Compression and Compensation
• Iterative Predictor-Critic Code Decoding for Real-World Image Dehazing
• Reconciling Stochastic and Deterministic Strategies for Zero-shot Image Restoration using Diffusion Model in Dual
• Complexity Experts are Task-Discriminative Learners for Any Image Restoration
• Adaptive Dropout: Unleashing Dropout across Layers for Generalizable Image Super-Resolution
• Linear Attention Modeling for Learned Image Compression
• SnowMaster: Comprehensive Real-world Image Desnowing via MLLM with Multi-Model Feedback Optimization
• Multi-Modal Synergistic Implicit Image Enhancement for Efficient Optical Flow Estimation
• Channel Consistency Prior and Self-Reconstruction Strategy Based Unsupervised Image Deraining
• Distilling Spatially-Heterogeneous Distortion Perception for Blind Image Quality Assessment
• PIDSR: Complementary Polarized Image Demosaicing and Super-Resolution
• A Polarization-Aided Transformer for Image Deblurring via Motion Vector Decomposition
• Neural Video Compression with Context Modulation
• Using Powerful Prior Knowledge of Diffusion Model in Deep Unfolding Networks for Image Compressive Sensing
• Improving Visual and Downstream Performance of Low-Light Enhancer with Vision Foundation Models Collaboration

总结：low-level比我想象中多，super-resolution应该是最多的，并且image compression貌似得到了比较大的关注。除此以外，去雾去雨去学ISP基础等每个方向只有1-2篇，有部分工作用LVM做low-level的问题，可以关注

其他

• Teaching Large Language Models to Regress Accurate Image Quality Scores Using Score Distribution

总结

出现最多的关键词： LVM\LLM，3DGS，image generation\understanding\editing\inpainting、multi-view，robotics
个人感受： LVM\LLM\robotics确实是大热点，LVM\LLM多做下游任务，有部分微调的工作。robotics上面的任务非常繁杂。
超出以外的是有非常多multi-view关键词的工作，个人猜想是否是因为2D CV的工作其实已经到一个好的阶段，但是总归要应用到3D中，因此探索multi-view：2D-3D-2D的过程是正在进行时
与大论文相关的工作： 出现了一些 Aerial-to-Ground关键词，并且有两篇是数据集贡献。