ICCV 2021 accepted papers on Detection and segmentation
222 of the 1,612 papers accepted at ICCV 2021 (13.8%) are on Detection and segmentation, compared with 18.6% at ICCV 2019. They fall into 5 topics.
See ICCV 2021 in the interactive map
Submitting to ICCV? Checklist →
Working on something in this area? Describe your idea in scime atlas to see the closest papers by meaning and how crowded the spot has become.
Topics and papers
CNN features · 96 papers
- Swin Transformer: Hierarchical Vision Transformer Using Shifted Windows (38,712 citations)
- Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction Without Convolutions (5,494 citations)
- Segmenter: Transformer for Semantic Segmentation (2,393 citations)
- TOOD: Task-Aligned One-Stage Object Detection (1,759 citations)
- FcaNet: Frequency Channel Attention Networks (1,614 citations)
- Oriented R-CNN for Object Detection (1,549 citations)
- Conditional DETR for Fast Training Convergence (1,130 citations)
- Conformer: Local Features Coupling Global Representations for Visual Recognition (934 citations)
- Exploring Cross-Image Pixel Contrast for Semantic Segmentation (618 citations)
- Pixel Difference Networks for Efficient Edge Detection (579 citations)
- Dynamic DETR: End-to-End Object Detection With Dynamic Attention (472 citations)
- Co-Scale Conv-Attentional Image Transformers (464 citations)
and 84 more in the interactive map.
Open-set and few-shot detection · 55 papers
- ACDC: The Adverse Conditions Dataset With Correspondences for Semantic Driving Scene Understanding (620 citations)
- Unsupervised Semantic Segmentation by Contrasting Object Mask Proposals (237 citations)
- Simpler Is Better: Few-Shot Semantic Segmentation With Classifier Weight Transformer (224 citations)
- Mining Latent Classes for Few-Shot Segmentation (167 citations)
- KoDF: A Large-Scale Korean DeepFake Detection Dataset (165 citations)
- Query Adaptive Few-Shot Object Detection With Heterogeneous Graph Convolutional Networks (158 citations)
- Recurrent Mask Refinement for Few-Shot Medical Image Segmentation (156 citations)
- Vector-Decomposed Disentanglement for Domain-Invariant Object Detection (154 citations)
- Learning Meta-Class Memory for Few-Shot Semantic Segmentation (150 citations)
- Spatially Conditioned Graphs for Detecting Human-Object Interactions (143 citations)
- Universal-Prototype Enhancing for Few-Shot Object Detection (131 citations)
- Exploring Classification Equilibrium in Long-Tailed Object Detection (118 citations)
and 43 more in the interactive map.
Weakly supervised detection and segmentation · 33 papers
- End-to-End Semi-Supervised Object Detection With Soft Teacher (615 citations)
- TS-CAM: Token Semantic Coupled Attention Map for Weakly Supervised Object Localization (244 citations)
- Contrastive Learning for Label Efficient Semantic Segmentation (195 citations)
- Change Is Everywhere: Single-Temporal Supervised Object Change Detection in Remote Sensing Imagery (191 citations)
- Leveraging Auxiliary Tasks With Affinity Learning for Weakly Supervised Semantic Segmentation (183 citations)
- Complementary Patch for Weakly Supervised Semantic Segmentation (161 citations)
- Unlocking the Potential of Ordinary Classifier: Class-Specific Adversarial Erasing Framework for Weakly Supervised Semantic Segmentation (158 citations)
- Context Decoupling Augmentation for Weakly Supervised Semantic Segmentation (150 citations)
- ECS-Net: Improving Weakly Supervised Semantic Segmentation by Using Connections Between Class Activation Maps (135 citations)
- DiscoBox: Weakly Supervised Instance Segmentation and Semantic Correspondence From Box Supervision (122 citations)
- MFNet: Multi-Filter Directive Network for Weakly Supervised Salient Object Detection (111 citations)
- Pseudo-Mask Matters in Weakly-Supervised Semantic Segmentation (109 citations)
and 21 more in the interactive map.
3D detection and driving · 20 papers
- An End-to-End Transformer Model for 3D Object Detection (532 citations)
- Is Pseudo-Lidar Needed for Monocular 3D Object Detection? (355 citations)
- Geometry Uncertainty Projection Network for Monocular 3D Object Detection (280 citations)
- RangeDet: In Defense of Range View for LiDAR-Based 3D Object Detection (267 citations)
- Geometry-Based Distance Decomposition for Monocular 3D Object Detection (170 citations)
- SPG: Unsupervised Domain Adaptation for 3D Object Detection via Semantic Point Generation (160 citations)
- AutoShape: Real-Time Shape-Aware Monocular 3D Object Detection (151 citations)
- LIGA-Stereo: Learning LiDAR Geometry Aware Representations for Stereo-Based 3D Detector (124 citations)
- Robust Small Object Detection on the Water Surface Through Fusion of Camera and Millimeter Wave Radar (110 citations)
- Multi-View Radar Semantic Segmentation (107 citations)
- Unsupervised Domain Adaptive 3D Detection With Multi-Level Consistency (83 citations)
- Exploring Geometry-Aware Contrast and Clustering Harmonization for Self-Supervised 3D Object Detection (66 citations)
and 8 more in the interactive map.
Tracking · 18 papers
- Learning Spatio-Temporal Transformer for Visual Tracking (1,348 citations)
- Learning Target Candidate Association To Keep Track of What Not To Track (413 citations)
- Learn To Match: Automatic Matching Network Design for Visual Tracking (273 citations)
- Learning To Track With Object Permanence (241 citations)
- High-Performance Discriminative Tracking With Transformers (155 citations)
- Saliency-Associated Object Tracking (152 citations)
- Object Tracking by Jointly Exploiting Frame and Event Domain (143 citations)
- MOTSynth: How Can Synthetic Data Help Pedestrian Detection and Tracking? (135 citations)
- DepthTrack: Unveiling the Power of RGBD Tracking (105 citations)
- Track Without Appearance: Learn Box and Tracklet Embedding With Local and Global Motion Patterns for Vehicle Tracking (78 citations)
- A General Recurrent Tracking Framework Without Real Data (70 citations)
- Learning To Adversarially Blur Visual Object Tracking (65 citations)
and 6 more in the interactive map.
