1. Bridging the Gap Between Anchor-based and Anchor-free Detection via Adaptive Training Sample Selection论文地址：https://arxiv.org/abs/1912.02424代码：https://github.com/sfzhang15/ATSS2. Few-Shot Object Detection with Attention-RPN and Multi-Relation Detector论文地址：https://arxiv.org/abs/1908.01998
图像分割
1. Semi-Supervised Semantic Image Segmentation with Self-correcting Networks论文地址：https://arxiv.org/abs/1811.070732. Deep Snake for Real-Time Instance Segmentation论文地址：https://arxiv.org/abs/2001.016293. CenterMask : Real-Time Anchor-Free Instance Segmentation论文地址：https://arxiv.org/abs/1911.06667代码：https://github.com/youngwanLEE/CenterMask4. SketchGCN: Semantic Sketch Segmentation with Graph Convolutional Networks论文地址：https://arxiv.org/abs/2003.006785. PolarMask: Single Shot Instance Segmentation with Polar Representation论文地址：https://arxiv.org/abs/1909.13226代码：https://github.com/xieenze/PolarMask
6. xMUDA: Cross-Modal Unsupervised Domain Adaptation for 3D Semantic Segmentation论文地址：https://arxiv.org/abs/1911.126767. BlendMask: Top-Down Meets Bottom-Up for Instance Segmentation论文地址：https://arxiv.org/abs/2001.00309
人脸识别
1. Towards Universal Representation Learning for Deep Face Recognition论文地址：https://arxiv.org/abs/2002.118412. Suppressing Uncertainties for Large-Scale Facial Expression Recognition 论文地址：https://arxiv.org/abs/2002.10392代码：https://github.com/kaiwang960112/Self-Cure-Network3.Face X-ray for More General Face Forgery Detection论文地址：https://arxiv.org/pdf/1912.13458.pdf
1. PF-Net: Point Fractal Network for 3D Point Cloud Completion论文地址：https://arxiv.org/abs/2003.004102. PointAugment: an Auto-Augmentation Framework for Point Cloud Classification论文地址：https://arxiv.org/abs/2002.10876代码：https://github.com/liruihui/PointAugment/3.Learning multiview 3D point cloud registration论文地址：https://arxiv.org/abs/2001.051194. C-Flow: Conditional Generative Flow Models for Images and 3D Point Clouds论文地址：https://arxiv.org/abs/1912.070095. RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Clouds论文地址：https://arxiv.org/abs/1911.11236
6. Total3DUnderstanding: Joint Layout, Object Pose and Mesh Reconstruction for Indoor Scenes from a Single Image论文地址：https://arxiv.org/abs/2002.122127. Implicit Functions in Feature Space for 3D Shape Reconstruction and Completion论文地址：https://arxiv.org/abs/2003.014568. In Perfect Shape: Certifiably Optimal 3D Shape Reconstruction from 2D Landmarks论文地址：https://arxiv.org/pdf/1911.11924.pdf
图像处理
1. Learning to Shade Hand-drawn Sketches论文地址：https://arxiv.org/abs/2002.118122.Single Image Reflection Removal through Cascaded Refinement论文地址：https://arxiv.org/abs/1911.066343.Generalized ODIN: Detecting Out-of-distribution Image without Learning from Out-of-distribution Data论文地址：https://arxiv.org/abs/2002.112974. Deep Image Harmonization via Domain Verification论文地址：https://arxiv.org/abs/1911.13239代码：https://github.com/bcmi/Image_Harmonization_Datasets5. RoutedFusion: Learning Real-time Depth Map Fusion论文地址：https://arxiv.org/pdf/2001.04388.pdf
图像分类
1. Self-training with Noisy Student improves ImageNet classification论文地址：https://arxiv.org/abs/1911.042522. Image Matching across Wide Baselines: From Paper to Practice论文地址：https://arxiv.org/abs/2003.015873. Towards Robust Image Classification Using Sequential Attention Models论文地址：https://arxiv.org/abs/1912.02184
姿态估计
1. VIBE: Video Inference for Human Body Pose and Shape Estimation论文地址：https://arxiv.org/abs/1912.05656代码：https://github.com/mkocabas/VIBE2. Distribution-Aware Coordinate Representation for Human Pose Estimation论文地址：https://arxiv.org/abs/1910.06278代码：https://github.com/ilovepose/DarkPose3. 4D Association Graph for Realtime Multi-person Motion Capture Using Multiple Video Cameras论文地址：https://arxiv.org/abs/2002.126254. Optimal least-squares solution to the hand-eye calibration problem论文地址：https://arxiv.org/abs/2002.108385. D3VO: Deep Depth, Deep Pose and Deep Uncertainty for Monocular Visual Odometry论文地址：https://arxiv.org/abs/2003.010606. Multi-Modal Domain Adaptation for Fine-Grained Action Recognition论文地址：https://arxiv.org/abs/2001.096917. Distribution Aware Coordinate Representation for Human Pose Estimation论文地址：https://arxiv.org/abs/1910.062788. The Devil is in the Details: Delving into Unbiased Data Processing for Human Pose Estimation论文地址：https://arxiv.org/abs/1911.075249.PVN3D: A Deep Point-wise 3D Keypoints Voting Network for 6DoF Pose Estimation论文地址：https://arxiv.org/abs/1911.04231
视频分析
1. Rethinking Zero-shot Video Classification: End-to-end Training for Realistic Applications论文地址：https://arxiv.org/abs/2003.01455代码：https://github.com/bbrattoli/ZeroShotVideoClassification2. Say As You Wish: Fine-grained Control of Image Caption Generation with Abstract Scene Graphs论文地址：https://arxiv.org/abs/2003.003873. Fine-grained Video-Text Retrieval with Hierarchical Graph Reasoning论文地址：https://arxiv.org/abs/2003.003924. Object Relational Graph with Teacher-Recommended Learning for Video Captioning论文地址：https://arxiv.org/abs/2002.115665. Zooming Slow-Mo: Fast and Accurate One-Stage Space-Time Video Super-Resolution论文地址：https://arxiv.org/abs/2002.116166. Blurry Video Frame Interpolation论文地址：https://arxiv.org/abs/2002.122597. Hierarchical Conditional Relation Networks for Video Question Answering论文地址：https://arxiv.org/abs/2002.106988. Action Modifiers:Learning from Adverbs in Instructional Video论文地址：https://arxiv.org/abs/1912.06617
OCR
1. ABCNet: Real-time Scene Text Spotting with Adaptive Bezier-Curve Network论文地址：https://arxiv.org/abs/2002.10200代码：https://github.com/Yuliang-Liu/bezier_curve_text_spotting,https://github.com/aim-uofa/adet
GAN
1. Your Local GAN: Designing Two Dimensional Local Attention Mechanisms for Generative Models论文地址：https://arxiv.org/abs/1911.12287代码：https://github.com/giannisdaras/ylg2. MSG-GAN: Multi-Scale Gradient GAN for Stable Image Synthesis论文地址：https://arxiv.org/abs/1903.060483. Robust Design of Deep Neural Networks against Adversarial Attacks based on Lyapunov Theory论文地址：https://arxiv.org/abs/1911.04636
小样本&零样本
1. Improved Few-Shot Visual Classification论文地址：https://arxiv.org/pdf/1912.03432.pdf2.Meta-Transfer Learning for Zero-Shot Super-Resolution论文地址：https://arxiv.org/abs/2002.12213
弱监督&无监督
1. Rethinking the Route Towards Weakly Supervised Object Localization论文地址：https://arxiv.org/abs/2002.11359
2. NestedVAE: Isolating Common Factors via Weak Supervision论文地址：https://arxiv.org/abs/2002.115763.Unsupervised Reinforcement Learning of Transferable Meta-Skills for Embodied Navigation论文地址：https://arxiv.org/abs/1911.074504.Disentangling Physical Dynamics from Unknown Factors for Unsupervised Video Prediction论文地址：https://arxiv.org/abs/2003.01460
神经网络
1. Visual Commonsense R-CNN论文地址：https://arxiv.org/abs/2002.122042. GhostNet: More Features from Cheap Operations论文地址：https://arxiv.org/abs/1911.11907代码：https://github.com/iamhankai/ghostnet3. Watch your Up-Convolution: CNN Based Generative Deep Neural Networks are Failing to Reproduce Spectral 论文地址：https://arxiv.org/abs/2003.01826
模型加速
1. GPU-Accelerated Mobile Multi-view Style Transfer论文地址：https://arxiv.org/abs/2003.00706
视觉常识
1. What it Thinks is Important is Important: Robustness Transfers through Input Gradients论文地址：https://arxiv.org/abs/1912.056992.Attentive Context Normalization for Robust Permutation-Equivariant Learning论文地址：https://arxiv.org/abs/1907.025453. Bundle Adjustment on a Graph Processor论文地址：https://arxiv.org/abs/2003.03134https://github.com/joeaortiz/gbp4. Transferring Dense Pose to Proximal Animal Classes论文地址：https://arxiv.org/abs/2003.000805. Representations, Metrics and Statistics For Shape Analysis of Elastic Graphs论文地址：https://arxiv.org/abs/2003.002876. Learning in the Frequency Domain论文地址：https://arxiv.org/abs/2002.124167.Filter Grafting for Deep Neural Networks论文地址：https://arxiv.org/pdf/2001.05868.pdf8.ClusterFit: Improving Generalization of Visual Representations论文地址：https://arxiv.org/abs/1912.033309.Social-STGCNN: A Social Spatio-Temporal Graph Convolutional Neural Network for Human Trajectory Prediction论文地址：https://arxiv.org/abs/2002.1192710.Auto-Encoding Twin-Bottleneck Hashing论文地址：https://arxiv.org/abs/2002.1193011.Learning Representations by Predicting Bags of Visual Words论文地址：https://arxiv.org/abs/2002.1224712.Holistically-Attracted Wireframe Parsing论文地址：https://arxiv.org/abs/2003.0166313.A General and Adaptive Robust Loss Function论文地址：https://arxiv.org/abs/1701.0307714.A Characteristic Function Approach to Deep Implicit Generative Modeling论文地址：https://arxiv.org/abs/1909.0742515.AdderNet: Do We Really Need Multiplications in Deep Learning? 论文地址：https://arxiv.org/pdf/1912.1320016.12-in-1: Multi-Task Vision and Language Representation Learning论文地址：https://arxiv.org/abs/1912.0231517.Making Better Mistakes: Leveraging Class Hierarchies with Deep Networks论文地址：https://arxiv.org/abs/1912.0939318.CARS: Contunuous Evolution for Efficient Neural Architecture Search论文地址：https://arxiv.org/pdf/1909.04977.pdf代码：https://github.com/huawei-noah/CARS19.Towards Learning a Generic Agent for Vision-and-Language Navigation via Pre-training论文地址：https://arxiv.org/abs/2002.10638代码：https://github.com/weituo12321/PREVALENT
