ECCV2022 &CVPR2022论文速递2022.7.20！& demo

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

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 : 20 Jul 2022

total number : 26

PoserNet: Refining Relative Camera Poses Exploiting Object Detections

• 论文/Paper: http://arxiv.org/pdf/2207.09445
• 代码/Code: https://github.com/IIT-PAVIS/PoserNet.

Geometric Features Informed Multi-person Human-object Interaction Recognition in Videos

• 论文/Paper: http://arxiv.org/pdf/2207.09425
• 代码/Code: None

RCLane: Relay Chain Prediction for Lane Detection

• 论文/Paper: http://arxiv.org/pdf/2207.09399
• 代码/Code: None

Rethinking IoU-based Optimization for Single-stage 3D Object Detection

• 论文/Paper: http://arxiv.org/pdf/2207.09332
• 代码/Code: https://github.com/hlsheng1/RDIoU

Deep Semantic Statistics Matching (D2SM) Denoising Network

• 论文/Paper: http://arxiv.org/pdf/2207.09302
• 代码/Code: None

The Caltech Fish Counting Dataset: A Benchmark for Multiple-Object Tracking and Counting

• 论文/Paper: http://arxiv.org/pdf/2207.09295
• 代码/Code: None

3D Room Layout Estimation from a Cubemap of Panorama Image via Deep Manhattan Hough Transform

• 论文/Paper: http://arxiv.org/pdf/2207.09291
• 代码/Code: https://github.com/Starrah/DMH-Net.

Action Quality Assessment with Temporal Parsing Transformer

• 论文/Paper: http://arxiv.org/pdf/2207.09270
• 代码/Code: None

Image Super-Resolution with Deep Dictionary

• 论文/Paper: http://arxiv.org/pdf/2207.09228
• 代码/Code: None

NDF: Neural Deformable Fields for Dynamic Human Modelling

• 论文/Paper: http://arxiv.org/pdf/2207.09193
• 代码/Code: None

Self-Supervision Can Be a Good Few-Shot Learner

• 论文/Paper: http://arxiv.org/pdf/2207.09176
• 代码/Code: https://github.com/bbbdylan/unisiam

Single Stage Virtual Try-on via Deformable Attention Flows

• 论文/Paper: http://arxiv.org/pdf/2207.09161
• 代码/Code: None

FedX: Unsupervised Federated Learning with Cross Knowledge Distillation

• 论文/Paper: http://arxiv.org/pdf/2207.09158
• 代码/Code: None

Learning Mutual Modulation for Self-Supervised Cross-Modal Super-Resolution

• 论文/Paper: http://arxiv.org/pdf/2207.09156
• 代码/Code: None

What Matters for 3D Scene Flow Network

• 论文/Paper: http://arxiv.org/pdf/2207.09143
• 代码/Code: https://github.com/IRMVLab/3DFlow.

ParticleSfM: Exploiting Dense Point Trajectories for Localizing Moving Cameras in the Wild

• 论文/Paper: http://arxiv.org/pdf/2207.09137
• 代码/Code: https://github.com/bytedance/particle-sfm.

MHR-Net: Multiple-Hypothesis Reconstruction of Non-Rigid Shapes from 2D Views

• 论文/Paper: http://arxiv.org/pdf/2207.09086
• 代码/Code: None

Dual Adaptive Transformations for Weakly Supervised Point Cloud Segmentation

• 论文/Paper: http://arxiv.org/pdf/2207.09084
• 代码/Code: None

Box-supervised Instance Segmentation with Level Set Evolution

• 论文/Paper: http://arxiv.org/pdf/2207.09055
• 代码/Code: https://github.com/LiWentomng/boxlevelset.

ML-BPM: Multi-teacher Learning with Bidirectional Photometric Mixing for Open Compound Domain Adaptation in Semantic Segmentation

• 论文/Paper: http://arxiv.org/pdf/2207.09045
• 代码/Code: None

Structure-aware Editable Morphable Model for 3D Facial Detail Animation and Manipulation

• 论文/Paper: http://arxiv.org/pdf/2207.09019
• 代码/Code: https://github.com/gerwang/facial-detail-manipulation.

SelectionConv: Convolutional Neural Networks for Non-rectilinear Image Data

• 论文/Paper: http://arxiv.org/pdf/2207.08979
• 代码/Code: None

Exploiting Unlabeled Data with Vision and Language Models for Object Detection

• 论文/Paper: http://arxiv.org/pdf/2207.08954
• 代码/Code: https://github.com/xiaofeng94/VL-PLM.

Prior-Guided Adversarial Initialization for Fast Adversarial Training

• 论文/Paper: http://arxiv.org/pdf/2207.08859
• 代码/Code: https://github.com/jiaxiaojunQAQ/FGSM-PGI.

Self-Supervised Interactive Object Segmentation Through a Singulation-and-Grasping Approach

• 论文/Paper: http://arxiv.org/pdf/2207.09314
• 代码/Code: None

Prior Knowledge Guided Unsupervised Domain Adaptation

• 论文/Paper: http://arxiv.org/pdf/2207.08877
• 代码/Code: https://github.com/tsun/KUDA.

CVPR2022

Updated on : 20 Jul 2022

total number : 2

Visual Representation Learning with Transformer: A Sequence-to-Sequence Perspective

• 论文/Paper: http://arxiv.org/pdf/2207.09339
• 代码/Code: None

Balanced Contrastive Learning for Long-Tailed Visual Recognition

• 论文/Paper: http://arxiv.org/pdf/2207.09052
• 代码/Code: \href{https://github.com/FlamieZhu/BCL}{this