搜索

ECCV2022 &CVPR2022论文速递2022.7.21！& demo

AI算法与图像处理

共 7342字，需浏览 15分钟

 ·

2022-07-22 12:23


整理：AI算法与图像处理
CVPR2022论文和代码整理：https://github.com/DWCTOD/CVPR2022-Papers-with-Code-Demo
ECCV2022论文和代码整理：https://github.com/DWCTOD/ECCV2022-Papers-with-Code-Demo
欢迎关注公众号 AI算法与图像处理，获取更多干货：


大家好,  最近正在优化每周分享的CVPR$ECCV 2022论文, 目前考虑按照不同类别去分类,方便不同方向的小伙伴挑选自己感兴趣的论文哈
欢迎大家留言其他想法,  合适的话会采纳哈! 求个三连支持一波哈

建了一个知识星球，计划不定期分享最新的成果和资源！感兴趣可以扫描体验，另外还有50个一年免费体验名额，可以添加微信nvshenj125 申请。

最新成果demo展示：


ECCV2022 | XMem: 高质量长期视频分割！

效果超群！


标题：XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model

论文：https://arxiv.org/pdf/2207.07115.pdf

代码：https://github.com/hkchengrex/XMem


摘要：

我们提出了 XMem，这是一种用于长视频的视频对象分割架构，具有统一的特征内存存储，受 Atkinson-Shiffrin 内存模型的启发。先前关于视频对象分割的工作通常只使用一种类型的特征记忆。对于超过一分钟的视频，单个特征内存模型将内存消耗和准确性紧密联系在一起。相比之下，遵循 Atkinson-Shiffrin 模型，我们开发了一种架构，该架构包含多个独立但深度连接的特征记忆存储：快速更新的感觉记忆、高分辨率工作记忆和紧凑的持续长期记忆。至关重要的是，我们开发了一种记忆增强算法，该算法通常将积极使用的工作记忆元素整合到长期记忆中，从而避免记忆爆炸并最大限度地减少长期预测的性能衰减。结合新的内存读取机制，XMem 在长视频数据集上的性能大大超过了最先进的性能，同时在短视频上与最先进的方法（不适用于长视频）相当数据集。





最新论文整理


   ECCV2022


Updated on : 21 Jul 2022

total number : 43

Discover and Mitigate Unknown Biases with Debiasing Alternate Networks

  • 论文/Paper: http://arxiv.org/pdf/2207.10077

  • 代码/Code: https://github.com/zhihengli-UR/DebiAN

Monocular 3D Object Reconstruction with GAN Inversion

  • 论文/Paper: http://arxiv.org/pdf/2207.10061

  • 代码/Code: https://github.com/junzhezhang/mesh-inversion

3D Clothed Human Reconstruction in the Wild

  • 论文/Paper: http://arxiv.org/pdf/2207.10053

  • 代码/Code: https://github.com/hygenie1228/clothwild_release

Densely Constrained Depth Estimator for Monocular 3D Object Detection

  • 论文/Paper: http://arxiv.org/pdf/2207.10047

  • 代码/Code: https://github.com/bravegroup/dcd

MOTCOM: The Multi-Object Tracking Dataset Complexity Metric

  • 论文/Paper: http://arxiv.org/pdf/2207.10031

  • 代码/Code: None

Difficulty-Aware Simulator for Open Set Recognition

  • 论文/Paper: http://arxiv.org/pdf/2207.10024

  • 代码/Code: https://github.com/wjun0830/difficulty-aware-simulator

Tailoring Self-Supervision for Supervised Learning

  • 论文/Paper: http://arxiv.org/pdf/2207.10023

  • 代码/Code: https://github.com/wjun0830/localizable-rotation

Generative Domain Adaptation for Face Anti-Spoofing

  • 论文/Paper: http://arxiv.org/pdf/2207.10015

  • 代码/Code: None

Overcoming Shortcut Learning in a Target Domain by Generalizing Basic Visual Factors from a Source Domain

  • 论文/Paper: http://arxiv.org/pdf/2207.10002

  • 代码/Code: https://github.com/boschresearch/sourcegen

DecoupleNet: Decoupled Network for Domain Adaptive Semantic Segmentation

  • 论文/Paper: http://arxiv.org/pdf/2207.09988

  • 代码/Code: https://github.com/dvlab-research/decouplenet

Temporal and cross-modal attention for audio-visual zero-shot learning

  • 论文/Paper: http://arxiv.org/pdf/2207.09966

  • 代码/Code: https://github.com/explainableml/tcaf-gzsl

Telepresence Video Quality Assessment

  • 论文/Paper: http://arxiv.org/pdf/2207.09956

  • 代码/Code: None

Learning Pedestrian Group Representations for Multi-modal Trajectory Prediction

  • 论文/Paper: http://arxiv.org/pdf/2207.09953

  • 代码/Code: https://github.com/inhwanbae/gpgraph

Towards Efficient and Scale-Robust Ultra-High-Definition Image Demoireing

  • 论文/Paper: http://arxiv.org/pdf/2207.09935

  • 代码/Code: None

Robust Landmark-based Stent Tracking in X-ray Fluoroscopy

  • 论文/Paper: http://arxiv.org/pdf/2207.09933

  • 代码/Code: None

Negative Samples are at Large: Leveraging Hard-distance Elastic Loss for Re-identification

  • 论文/Paper: http://arxiv.org/pdf/2207.09884

  • 代码/Code: None

Discrete-Constrained Regression for Local Counting Models

  • 论文/Paper: http://arxiv.org/pdf/2207.09865

  • 代码/Code: None

UNIF: United Neural Implicit Functions for Clothed Human Reconstruction and Animation

  • 论文/Paper: http://arxiv.org/pdf/2207.09835

  • 代码/Code: https://github.com/ShenhanQian/UNIF

CoSMix: Compositional Semantic Mix for Domain Adaptation in 3D LiDAR Segmentation

  • 论文/Paper: http://arxiv.org/pdf/2207.09778

  • 代码/Code: https://github.com/saltoricristiano/cosmix-uda

Collaborating Domain-shared and Target-specific Feature Clustering for Cross-domain 3D Action Recognition

  • 论文/Paper: http://arxiv.org/pdf/2207.09767

  • 代码/Code: https://github.com/canbaoburen/CoDT

GIPSO: Geometrically Informed Propagation for Online Adaptation in 3D LiDAR Segmentation

  • 论文/Paper: http://arxiv.org/pdf/2207.09763

  • 代码/Code: https://github.com/saltoricristiano/gipso-sfouda

Feature Representation Learning for Unsupervised Cross-domain Image Retrieval

  • 论文/Paper: http://arxiv.org/pdf/2207.09721

  • 代码/Code: https://github.com/conghuihu/ucdir

Resolving Copycat Problems in Visual Imitation Learning via Residual Action Prediction

  • 论文/Paper: http://arxiv.org/pdf/2207.09705

  • 代码/Code: None

Robust Object Detection With Inaccurate Bounding Boxes

  • 论文/Paper: http://arxiv.org/pdf/2207.09697

  • 代码/Code: https://github.com/cxliu0/OA-MIL.

Efficient Meta-Tuning for Content-aware Neural Video Delivery

  • 论文/Paper: http://arxiv.org/pdf/2207.09691

  • 代码/Code: https://github.com/neural-video-delivery/emt-pytorch-eccv2022

Object-Compositional Neural Implicit Surfaces

  • 论文/Paper: http://arxiv.org/pdf/2207.09686

  • 代码/Code: https://github.com/qianyiwu/objsdf

Explaining Deepfake Detection by Analysing Image Matching

  • 论文/Paper: http://arxiv.org/pdf/2207.09679

  • 代码/Code: https://github.com/megvii-research/fst-matching

ERA: Expert Retrieval and Assembly for Early Action Prediction

  • 论文/Paper: http://arxiv.org/pdf/2207.09675

  • 代码/Code: None

GRIT: Faster and Better Image captioning Transformer Using Dual Visual Features

  • 论文/Paper: http://arxiv.org/pdf/2207.09666

  • 代码/Code: https://github.com/davidnvq/grit

Streamable Neural Fields

  • 论文/Paper: http://arxiv.org/pdf/2207.09663

  • 代码/Code: https://github.com/jwcho5576/streamable_nf

Unsupervised Domain Adaptation for One-stage Object Detector using Offsets to Bounding Box

  • 论文/Paper: http://arxiv.org/pdf/2207.09656

  • 代码/Code: None

Learning Topological Interactions for Multi-Class Medical Image Segmentation

  • 论文/Paper: http://arxiv.org/pdf/2207.09654

  • 代码/Code: https://github.com/topoxlab/topointeraction

Aware of the History: Trajectory Forecasting with the Local Behavior Data

  • 论文/Paper: http://arxiv.org/pdf/2207.09646

  • 代码/Code: None

Hierarchically Self-Supervised Transformer for Human Skeleton Representation Learning

  • 论文/Paper: http://arxiv.org/pdf/2207.09644

  • 代码/Code: None

Perspective Phase Angle Model for Polarimetric 3D Reconstruction

  • 论文/Paper: http://arxiv.org/pdf/2207.09629

  • 代码/Code: https://github.com/gcchen97/ppa4p3d

Explicit Image Caption Editing

  • 论文/Paper: http://arxiv.org/pdf/2207.09625

  • 代码/Code: https://github.com/baaaad/ece

Unsupervised Deep Multi-Shape Matching

  • 论文/Paper: http://arxiv.org/pdf/2207.09610

  • 代码/Code: None

AiATrack: Attention in Attention for Transformer Visual Tracking

  • 论文/Paper: http://arxiv.org/pdf/2207.09603

  • 代码/Code: https://github.com/Little-Podi/AiATrack

Tip-Adapter: Training-free Adaption of CLIP for Few-shot Classification

  • 论文/Paper: http://arxiv.org/pdf/2207.09519

  • 代码/Code: https://github.com/gaopengcuhk/tip-adapter

Contributions of Shape, Texture, and Color in Visual Recognition

  • 论文/Paper: http://arxiv.org/pdf/2207.09510

  • 代码/Code: https://github.com/gyhandy/humanoid-vision-engine

An Efficient Method for Face Quality Assessment on the Edge

  • 论文/Paper: http://arxiv.org/pdf/2207.09505

  • 代码/Code: None

Invariant Feature Learning for Generalized Long-Tailed Classification

  • 论文/Paper: http://arxiv.org/pdf/2207.09504

  • 代码/Code: https://github.com/kaihuatang/generalized-long-tailed-benchmarks.pytorch

Single Frame Atmospheric Turbulence Mitigation: A Benchmark Study and A New Physics-Inspired Transformer Model

  • 论文/Paper: http://arxiv.org/pdf/2207.10040

  • 代码/Code: None


   CVPR2022




浏览 135
点赞
评论
收藏
分享

手机扫一扫分享

分享
举报
评论
图片
表情
推荐
点赞
评论
收藏
分享

手机扫一扫分享

分享
举报