Computer vision - ECCV 2022 . Part XVIII : 17th European Conference, Tel Aviv, Israel, October 23-27, 2022 : proceedings / / Shai Avidan [and four others]

• Pubblicazione: Cham, Switzerland : , : Springer, , [2022]; ©2022
• Descrizione fisica: 1 online resource (812 pages)
• Disciplina: 006.37
• Soggetto topico: Computer vision; Pattern recognition systems
• Persona (resp. second.): AvidanShai
• Nota di contenuto: Intro -- Foreword -- Preface -- Organization -- Contents - Part XVIII -- Dynamic Dual Trainable Bounds for Ultra-low Precision Super-Resolution Networks -- 1 Introduction -- 2 Related Work -- 2.1 Single Image Super Resolution -- 2.2 Quantized SR Models -- 3 Methodology -- 3.1 Preliminaries -- 3.2 Our Insights -- 3.3 Our Solutions -- 3.4 Training Loss -- 4 Experiments -- 4.1 Implementation Details -- 4.2 Experimental Results -- 4.3 Model Analysis -- 4.4 Ablation Study -- 5 Conclusion -- References -- OSFormer: One-Stage Camouflaged Instance Segmentation with Transformers -- 1 Introduction -- 2 Related Work -- 3 OSFormer -- 3.1 CNN Backbone -- 3.2 Location-Sensing Transformer -- 3.3 Coarse-to-Fine Fusion -- 3.4 Dynamic Camouflaged Instance Normalization -- 3.5 Loss Function -- 4 Experiments -- 4.1 Experimental Settings -- 4.2 Ablation Studies -- 4.3 Comparisons with Cutting-Edge Methods -- 5 Conclusion -- References -- Highly Accurate Dichotomous Image Segmentation -- 1 Introduction -- 2 Related Work -- 3 Proposed DIS5K Dataset -- 3.1 Data Collection and Annotation -- 3.2 Data Analysis -- 3.3 Dataset Splitting -- 4 Proposed IS-Net Baseline -- 4.1 Intermediate Supervision -- 5 Proposed HCE Metric -- 6 DIS5K Benchmark -- 6.1 Quantitative Evaluation -- 6.2 Qualitative Evaluation -- 6.3 Ablation Study -- 7 Conclusions -- References -- Boosting Supervised Dehazing Methods via Bi-level Patch Reweighting -- 1 Introduction -- 2 Related Work -- 3 Bi-level Dehazing Framework -- 3.1 Patch Reweighting -- 3.2 Bi-level Dehazing Framework -- 3.3 Relationship with RL -- 4 Experimental Results -- 4.1 Training, Validation and Testing Dataset -- 4.2 Implementation Details -- 4.3 Performance Results and Ablation Analysis -- 4.4 Patch Reweighting and Additional Experiment -- 5 Conclusion -- References -- Flow-Guided Transformer for Video Inpainting -- 1 Introduction. ; 2 Related Work -- 3 Method -- 3.1 Problem Formulation -- 3.2 Network Overview -- 3.3 Local Aggregation Flow Completion Network -- 3.4 Flow-Guided Transformer for Video Inpainting -- 4 Experiments -- 4.1 Settings -- 4.2 Implementation Details -- 4.3 Quantitative Evaluation -- 4.4 Qualitative Comparisons -- 4.5 Ablation Studies -- 5 Conclusion -- References -- Shift-Tolerant Perceptual Similarity Metric -- 1 Introduction -- 2 Related Work -- 3 Human Perception of Small Shifts -- 4 Effect of Small Shifts on Similarity Metrics -- 5 Elements of Shift-Tolerant Metrics -- 6 Experiments -- 7 Conclusion -- References -- Perception-Distortion Balanced ADMM Optimization for Single-Image Super-Resolution -- 1 Introduction -- 2 Related Work -- 2.1 Objective Quality versus Perceptual Quality -- 2.2 Perception-Distortion Trade-off -- 3 Method -- 3.1 Revisiting the Multi-objective SR Formulation -- 3.2 Low-Frequency Constrained SR (LFc-SR) -- 3.3 Alternating Optimization with ADMM -- 4 Experiments -- 4.1 Settings -- 4.2 Comparison with State-of-the-Art Methods -- 4.3 Ablation Studies -- 5 Conclusion and Limitations -- References -- VQFR: Blind Face Restoration with Vector-Quantized Dictionary and Parallel Decoder -- 1 Introduction -- 2 Related Work -- 3 Methodology -- 3.1 Vector-Quantized Codebook -- 3.2 Parallel Decoder -- 3.3 Model Objective -- 4 Experiments -- 4.1 Implementation and Evaluation Settings -- 4.2 Comparisons with State-of-the-art Methods -- 4.3 Ablation Study -- 5 Conclusion -- References -- Uncertainty Learning in Kernel Estimation for Multi-stage Blind Image Super-Resolution -- 1 Introduction -- 2 Related Work -- 2.1 Blind Image Super-Resolution -- 2.2 Uncertainty in Deep Learning -- 2.3 LSM Model for Image Restoration -- 3 Method -- 3.1 Problem Formulation -- 3.2 Uncertainty Learning in Kernel Estimation -- 3.3 Multi-stage SR Network. ; 3.4 Network Training -- 4 Experimental Results -- 4.1 Datasets and Settings -- 4.2 Comparing UL with Deterministic Network -- 4.3 Comparison with State-of-the-Art Methods -- 4.4 Results on Real-World Images -- 4.5 Ablation Study -- 5 Conclusions -- References -- Learning Spatio-Temporal Downsampling for Effective Video Upscaling -- 1 Introduction -- 2 Related Works -- 2.1 Video Downsampling -- 2.2 Video Upscaling -- 3 Space-Time Anti-Aliasing (STAA) -- 3.1 Intuition -- 3.2 Module Design -- 4 Joint Downscaling and Upscaling Framework -- 4.1 Downsampler -- 4.2 Upsampler -- 5 Experiments -- 5.1 Comparison with State-of-the-Art Methods -- 5.2 Ablation Studies -- 5.3 Applications -- 6 Conclusions -- References -- Learning Local Implicit Fourier Representation for Image Warping -- 1 Introduction -- 2 Related Works -- 3 Problem Formulation -- 3.1 Learning Fourier Information for Local Neural Representation -- 3.2 Learning Fourier Information with Coordinate Transformations -- 3.3 Shape-Dependent Phase Estimation -- 4 Methods -- 4.1 Architecture Details -- 4.2 Training Strategy -- 5 Experiments -- 5.1 Dataset and Training -- 5.2 Evaluation -- 5.3 Ablation Study -- 5.4 Fourier Feature Space -- 5.5 Discussion -- 6 Conclusions -- References -- SepLUT: Separable Image-Adaptive Lookup Tables for Real-Time Image Enhancement -- 1 Introduction -- 2 Related Works -- 2.1 Lookup Tables -- 2.2 Image Enhancement -- 3 Methods -- 3.1 Overall Framework -- 3.2 Global Image Context Analysis: Backbone Network -- 3.3 Component-Independent Transform: 1D Lookup Tables -- 3.4 Component-Correlated Transform: 3D Lookup Tables -- 3.5 Efficient Implementation via Quantization -- 4 Experiments -- 4.1 Datasets -- 4.2 Implementation Details -- 4.3 Ablation Studies -- 4.4 Analysis -- 4.5 Comparisons with State-of-the-Arts -- 5 Conclusion -- References -- Blind Image Decomposition. ; 1 Introduction -- 2 Related Work -- 3 Blind Image Decomposition Formulation -- 4 Blind Image Decomposition Network -- 4.1 Objective -- 4.2 Training Details -- 5 Blind Image Decomposition Tasks -- 5.1 Task I: Mixed Image Decomposition Across Multiple Domains -- 5.2 Task II: Real-Scenario Deraining -- 5.3 Task III: Joint Shadow/Reflection/Watermark Removal -- 6 Ablation Study and Analysis -- 7 Conclusion -- References -- MuLUT: Cooperating Multiple Look-Up Tables for Efficient Image Super-Resolution -- 1 Introduction -- 2 Related Works -- 3 Cooperation of Multiple LUTs for SR -- 3.1 Preliminary -- 3.2 Overview -- 3.3 Parallelizing LUTs with Complementary Indexing -- 3.4 Cascading LUTs with Hierarchical Indexing -- 3.5 The LUT-Aware Finetuning Strategy -- 4 Extension of MuLUT to Demosaicing -- 5 Experiments and Results -- 5.1 Experimental Settings -- 5.2 Quantitative Evaluation -- 5.3 Qualitative Evaluation -- 5.4 Ablation Studies -- 5.5 Results in Image Demosaicing -- 6 Conclusion Remarks -- References -- Learning Spatiotemporal Frequency-Transformer for Compressed Video Super-Resolution -- 1 Introduction -- 2 Related Work -- 2.1 Video Super-Resolution -- 2.2 Frequency Learning -- 3 Approach -- 3.1 Problem Formulation -- 3.2 Frequency-Based Tokenization -- 3.3 Frequency-Based Attention -- 3.4 Frequency Transformer -- 4 Experiments -- 4.1 Implementation Details -- 4.2 Datasets and Evaluation Metrics -- 4.3 Comparison with State-of-the-Art Methods -- 4.4 Ablation Study -- 5 Conclusions -- References -- Spatial-Frequency Domain Information Integration for Pan-Sharpening -- 1 Introduction -- 2 Related Work -- 2.1 Traditional Pan-Sharpening Methods -- 2.2 CNN-Based Pan-Sharpening Methods -- 3 Method -- 3.1 Fourier Transformation of Images -- 3.2 Framework -- 3.3 The Core Building Block -- 3.4 Loss Function -- 4 Experiments -- 4.1 Baseline Methods. ; 4.2 Implementation Details -- 4.3 Comparison with State-of-the-Art Methods -- 4.4 Parameter Numbers vs Model Performance -- 4.5 Ablation Experiments -- 4.6 Visualization of Feature Maps in Dual Domains -- 5 Conclusion -- References -- Adaptive Patch Exiting for Scalable Single Image Super-Resolution -- 1 Introduction -- 2 Related Work -- 3 Method -- 3.1 Training Multi-exit SR Networks -- 3.2 Estimating Incremental Capacity -- 3.3 Jointly Training SR Network and Regressor -- 4 Experiments -- 4.1 Implementation Details -- 4.2 Evaluation of APE -- 4.3 Ablation Study -- 4.4 Comparison with ClassSR and AdaDSR -- 5 Future Work -- 6 Conclusion -- References -- Efficient Meta-Tuning for Content-Aware Neural Video Delivery -- 1 Introduction -- 2 Related Work -- 3 Method -- 3.1 Overview -- 3.2 Efficient Meta-Tuning -- 3.3 Challenging Patch Sampling -- 4 Experiments -- 4.1 Experimental Details -- 4.2 Comparison with Baseline and Codec Standards -- 4.3 Comparison with Neural Video Delivery Methods -- 4.4 Ablation Study -- 4.5 The Generalization of Our Method -- 5 Extension to Long Videos -- 6 Conclusion -- References -- Reference-Based Image Super-Resolution with Deformable Attention Transformer -- 1 Introduction -- 2 Related Work -- 3 Proposed Method -- 3.1 Texture Feature Encoders -- 3.2 Reference-Based Deformable Attention -- 3.3 Residual Feature Aggregation -- 3.4 Loss Function -- 4 Experiments -- 4.1 Comparison with State-of-the-Art Methods -- 4.2 Further Analyses -- 4.3 More Evaluation Results -- 4.4 Discussion on Model Size -- 4.5 Ablation Study -- 5 Conclusion -- References -- Local Color Distributions Prior for Image Enhancement -- 1 Introduction -- 2 Related Work -- 3 Proposed Dataset -- 4 Proposed Method -- 4.1 Local Color Distribution (LCD) Pyramid -- 4.2 Proposed Network -- 4.3 Loss Function -- 5 Experiments -- 5.1 Implementation Details. ; 5.2 Comparisons with State-of-the-Art Methods.
• Titolo autorizzato: Computer Vision – ECCV 2022
• ISBN: 3-031-19797-6
• Formato: Materiale a stampa
• Livello bibliografico: Monografia
• Lingua di pubblicazione: Inglese
• Record Nr.: 9910629284603321
• Lo trovi qui: Univ. Federico II
• Serie: Lecture notes in computer science.