本文首发极市平台公众号,转载请获得授权并标明出处。 ECCV 2022 论文分方向整理目前在极市社区持续更新中,已累计更新了 54 篇,项目地址:https://github.com/extreme-assistant/ECCV2022-Paper-Code-Interpretation 以下是本周更新的 ECCV 2022 论文,包含检测,分割,图像处理,视频理解,神经网络结构设计,无监督学习,迁移学习等方向。 论文合集打包下载地址:https://www.cvmart.net/community/detail/6592 - 检测 - 分割 - 图像处理 - 视频处理 - 图像、视频检索与理解 - 估计 - 目标跟踪 - 文本检测与识别 - GAN/生成式/对抗式 - 神经网络结构设计 - 数据处理 - 模型训练/泛化 - 模型压缩 - 模型评估 - 半监督学习/自监督学习 - 多模态/跨模态学习 - 小样本学习 - 强化学习 检测2D目标检测 [1] Point-to-Box Network for Accurate Object Detection via Single Point Supervision (通过单点监督实现精确目标检测的点对盒网络) paper:https://arxiv.org/abs/2207.06827 code:https://github.com/ucas-vg/p2bnet [2] You Should Look at All Objects (您应该查看所有物体) paper:https://arxiv.org/abs/2207.07889 code:https://github.com/charlespikachu/yslao [3] Adversarially-Aware Robust Object Detector (对抗性感知鲁棒目标检测器) paper:https://arxiv.org/abs/2207.06202 code:https://github.com/7eu7d7/robustdet 3D目标检测 [1] Rethinking IoU-based Optimization for Single-stage 3D Object Detection (重新思考基于 IoU 的单阶段 3D 对象检测优化) paper:https://arxiv.org/abs/2207.09332 人物交互检测 [1] Towards Hard-Positive Query Mining for DETR-based Human-Object Interaction Detection (面向基于 DETR 的人机交互检测的硬性查询挖掘) paper:https://arxiv.org/abs/2207.05293 code:https://github.com/muchhair/hqm 图像异常检测 [1] DICE: Leveraging Sparsification for Out-of-Distribution Detection (DICE:利用稀疏化进行分布外检测) paper:https://arxiv.org/abs/2111.09805 code:https://github.com/deeplearning-wisc/dice 分割实例分割 [1] Box-supervised Instance Segmentation with Level Set Evolution (具有水平集进化的框监督实例分割) paper:https://arxiv.org/abs/2207.09055 [2] OSFormer: One-Stage Camouflaged Instance Segmentation with Transformers (OSFormer:使用 Transformers 进行单阶段伪装实例分割) paper:https://arxiv.org/abs/2207.02255 code:https://github.com/pjlallen/osformer 语义分割 [1] 2DPASS: 2D Priors Assisted Semantic Segmentation on LiDAR Point Clouds (2DPASS:激光雷达点云上的二维先验辅助语义分割) paper:https://arxiv.org/abs/2207.04397 code:https://github.com/yanx27/2dpass 视频目标分割 [1] Learning Quality-aware Dynamic Memory for Video Object Segmentation (视频对象分割的学习质量感知动态内存) paper:https://arxiv.org/abs/2207.07922 code:https://github.com/workforai/qdmn 图像处理超分辨率 [1] Dynamic Dual Trainable Bounds for Ultra-low Precision Super-Resolution Networks (超低精度超分辨率网络的动态双可训练边界) paper:https://arxiv.org/abs/2203.03844 code:https://github.com/zysxmu/ddtb 图像去噪 [1] Deep Semantic Statistics Matching (D2SM) Denoising Network (深度语义统计匹配(D2SM)去噪网络) paper:https://arxiv.org/abs/2207.09302 图像复原/图像增强/图像重建 [1] Semantic-Sparse Colorization Network for Deep Exemplar-based Colorization (用于基于深度示例的着色的语义稀疏着色网络) paper:https://arxiv.org/abs/2112.01335 [2] Geometry-aware Single-image Full-body Human Relighting (几何感知单图像全身人体重新照明) paper:https://arxiv.org/abs/2207.04750 [3] Multi-Modal Masked Pre-Training for Monocular Panoramic Depth Completion (单目全景深度补全的多模态蒙面预训练) paper:https://arxiv.org/abs/2203.09855 [4] PanoFormer: Panorama Transformer for Indoor 360 Depth Estimation (PanoFormer:用于室内 360 深度估计的全景变压器) paper:https://arxiv.org/abs/2203.09283 [5] SESS: Saliency Enhancing with Scaling and Sliding (SESS:通过缩放和滑动增强显着性) paper:https://arxiv.org/abs/2207.01769 [6] RigNet: Repetitive Image Guided Network for Depth Completion (RigNet:用于深度补全的重复图像引导网络) paper:https://arxiv.org/abs/2107.13802 图像外推(Image Outpainting) [1] Outpainting by Queries (通过查询进行外包) paper:https://arxiv.org/abs/2207.05312 code:https://github.com/kaiseem/queryotr 风格迁移(Style Transfer) [1] CCPL: Contrastive Coherence Preserving Loss for Versatile Style Transfer (CCPL:通用风格迁移的对比相干性保留损失) paper:https://arxiv.org/abs/2207.04808 code:https://github.com/JarrentWu1031/CCPL 视频处理(Video Processing) [1] Improving the Perceptual Quality of 2D Animation Interpolation (提高二维动画插值的感知质量) paper:https://arxiv.org/abs/2111.12792 code:https://github.com/shuhongchen/eisai-anime-interpolator [2] Real-Time Intermediate Flow Estimation for Video Frame Interpolation (视频帧插值的实时中间流估计) paper:https://arxiv.org/abs/2011.06294 code:https://github.com/MegEngine/arXiv2020-RIFE 图像、视频检索与理解动作识别 [1] ReAct: Temporal Action Detection with Relational Queries (ReAct:使用关系查询的时间动作检测) paper:https://arxiv.org/abs/2207.07097 code:https://github.com/sssste/react [2] Hunting Group Clues with Transformers for Social Group Activity Recognition (用Transformers寻找群体线索用于社会群体活动识别) paper:https://arxiv.org/abs/2207.05254 视频理解 [1] GraphVid: It Only Takes a Few Nodes to Understand a Video (GraphVid:只需几个节点即可理解视频) paper:https://arxiv.org/abs/2207.01375 [2] Deep Hash Distillation for Image Retrieval (用于图像检索的深度哈希蒸馏) paper:https://arxiv.org/abs/2112.08816 code:https://github.com/youngkyunjang/deep-hash-distillation 视频检索(Video Retrieval) [1] TS2-Net: Token Shift and Selection Transformer for Text-Video Retrieval (TS2-Net:用于文本视频检索的令牌移位和选择转换器) paper:https://arxiv.org/abs/2207.07852 code:https://github.com/yuqi657/ts2_net [2] Lightweight Attentional Feature Fusion: A New Baseline for Text-to-Video Retrieval (轻量级注意力特征融合:文本到视频检索的新基线) paper:https://arxiv.org/abs/2112.01832 估计位姿估计 [1] Category-Level 6D Object Pose and Size Estimation using Self-Supervised Deep Prior Deformation Networks (使用自监督深度先验变形网络的类别级 6D 对象姿势和大小估计) paper:https://arxiv.org/abs/2207.05444 code:https://github.com/jiehonglin/self-dpdn 深度估计 [1] Physical Attack on Monocular Depth Estimation with Optimal Adversarial Patches (使用最优对抗补丁对单目深度估计进行物理攻击) paper:https://arxiv.org/abs/2207.04718 目标跟踪 [1] Towards Grand Unification of Object Tracking (迈向目标跟踪的大统一) paper:https://arxiv.org/abs/2207.07078 code:https://github.com/masterbin-iiau/unicorn 文本检测与识别 [1] Dynamic Low-Resolution Distillation for Cost-Efficient End-to-End Text Spotting (用于经济高效的端到端文本识别的动态低分辨率蒸馏) paper:https://arxiv.org/abs/2207.06694 code:https://github.com/hikopensource/davar-lab-ocr GAN/生成式/对抗式 [1] Eliminating Gradient Conflict in Reference-based Line-Art Colorization (消除基于参考的艺术线条着色中的梯度冲突) paper:https://arxiv.org/abs/2207.06095 code:https://github.com/kunkun0w0/sga [2] WaveGAN: Frequency-aware GAN for High-Fidelity Few-shot Image Generation (WaveGAN:用于高保真少镜头图像生成的频率感知 GAN) paper:https://arxiv.org/abs/2207.07288 code:https://github.com/kobeshegu/eccv2022_wavegan [3] FakeCLR: Exploring Contrastive Learning for Solving Latent Discontinuity in Data-Efficient GANs (FakeCLR:探索对比学习以解决数据高效 GAN 中的潜在不连续性) paper:https://arxiv.org/abs/2207.08630 code:https://github.com/iceli1007/fakeclr [4] UniCR: Universally Approximated Certified Robustness via Randomized Smoothing (UniCR:通过随机平滑获得普遍近似的认证鲁棒性) paper:https://arxiv.org/abs/2207.02152 神经网络结构设计神经网络架构搜索(NAS) [1] ScaleNet: Searching for the Model to Scale (ScaleNet:搜索要扩展的模型) paper:https://arxiv.org/abs/2207.07267 code:https://github.com/luminolx/scalenet [2] Ensemble Knowledge Guided Sub-network Search and Fine-tuning for Filter Pruning (集成知识引导的子网络搜索和过滤器修剪微调) paper:https://arxiv.org/abs/2203.02651 code:https://github.com/sseung0703/ekg [3] EAGAN: Efficient Two-stage Evolutionary Architecture Search for GANs (EAGAN:GAN 的高效两阶段进化架构搜索) paper:https://arxiv.org/abs/2111.15097 code:https://github.com/marsggbo/EAGAN 数据处理归一化 [1] Fine-grained Data Distribution Alignment for Post-Training Quantization (训练后量化的细粒度数据分布对齐) paper:https://arxiv.org/abs/2109.04186 code:https://github.com/zysxmu/fdda 模型训练/泛化噪声标签 [1] Learning with Noisy Labels by Efficient Transition Matrix Estimation to Combat Label Miscorrection (通过有效的转移矩阵估计学习噪声标签以对抗标签错误校正) paper:https://arxiv.org/abs/2111.14932 模型压缩知识蒸馏 [1] Knowledge Condensation Distillation (知识浓缩蒸馏) paper:https://arxiv.org/abs/2207.05409 code:https://github.com/dzy3/kcd) 模型评估 [1] Hierarchical Latent Structure for Multi-Modal Vehicle Trajectory Forecasting (多模式车辆轨迹预测的分层潜在结构) paper:https://arxiv.org/abs/2207.04624 code:https://github.com/d1024choi/hlstrajforecast 半监督学习/无监督学习/自监督学习 [1] FedX: Unsupervised Federated Learning with Cross Knowledge Distillation (FedX:具有交叉知识蒸馏的无监督联合学习) paper:https://arxiv.org/abs/2207.09158 [2] Synergistic Self-supervised and Quantization Learning (协同自监督和量化学习) paper:https://arxiv.org/abs/2207.05432 code:https://github.com/megvii-research/ssql-eccv2022) [3] Contrastive Deep Supervision (对比深度监督) paper:https://arxiv.org/abs/2207.05306 code:https://github.com/archiplab-linfengzhang/contrastive-deep-supervision [4] Dense Teacher: Dense Pseudo-Labels for Semi-supervised Object Detection (稠密教师:用于半监督目标检测的稠密伪标签) paper:https://arxiv.org/abs/2207.02541 [5] Image Coding for Machines with Omnipotent Feature Learning (具有全能特征学习的机器的图像编码) paper:https://arxiv.org/abs/2207.01932 多模态学习/跨模态视觉-语言 [1] Contrastive Vision-Language Pre-training with Limited Resources (资源有限的对比视觉语言预训练) paper:https://arxiv.org/abs/2112.09331 code:https://github.com/zerovl/zerovl 跨模态 [1] Cross-modal Prototype Driven Network for Radiology Report Generation (用于放射学报告生成的跨模式原型驱动网络) paper:https://arxiv.org/abs/ code:https://github.com/markin-wang/xpronet 小样本学习 [1] Learning Instance and Task-Aware Dynamic Kernels for Few Shot Learning (用于少数镜头学习的学习实例和任务感知动态内核) paper:https://arxiv.org/abs/2112.03494 迁移学习/自适应 [1] Factorizing Knowledge in Neural Networks (在神经网络中分解知识) paper:https://arxiv.org/abs/2207.03337 code:https://github.com/adamdad/knowledgefactor [2] CycDA: Unsupervised Cycle Domain Adaptation from Image to Video (CycDA:从图像到视频的无监督循环域自适应) paper:https://arxiv.org/abs/2203.16244 强化学习 [1] Target-absent Human Attention (目标缺失——人类注意力缺失) paper:https://arxiv.org/abs/2207.01166 code:https://github.com/neouyghur/sess