精华置顶

墙裂推荐！小白如何1个月系统学习CV核心知识：链接

点击@CV计算机视觉，关注更多CV干货

论文已打包，点击进入—>下载界面

点击加入—>CV计算机视觉交流群

1.【基础网络架构】What Do Deep Saliency Models Learn about Visual Attention?

• 论文地址：https://arxiv.org//pdf/2310.09679

• 开源代码：GitHub - szzexpoi/saliency_analysis

2.【基础网络架构】KAKURENBO: Adaptively Hiding Samples in Deep Neural Network Training

• 论文地址：https://arxiv.org//pdf/2310.10102

• 开源代码（即将开源）：GitHub - TruongThaoNguyen/kakurenbo

3.【基础网络架构：CNN】RefConv: Re-parameterized Refocusing Convolution for Powerful ConvNets

• 论文地址：https://arxiv.org//pdf/2310.10563

• 开源代码：GitHub - Aiolus-X/RefConv

4.【图像分类】Hawkeye: A PyTorch-based Library for Fine-Grained Image Recognition with Deep Learning

• 论文地址：https://arxiv.org//pdf/2310.09600

• 开源代码：GitHub - Hawkeye-FineGrained/Hawkeye: Open source deep learning based fine-grained image recognition toolbox built on PyTorch🔥

5.【显著目标检测】Towards Open-World Co-Salient Object Detection with Generative Uncertainty-aware Group Selective Exchange-Masking

• 论文地址：https://arxiv.org//pdf/2310.10264

• 开源代码：GitHub - wuyang98/CoSOD

6.【图像分割】Label-efficient Segmentation via Affinity Propagation

• 论文地址：https://arxiv.org//pdf/2310.10533

• 工程主页：Label-efficient Segmentation via Affinity Propagation

• 开源代码：GitHub - CircleRadon/APro: The code for "Label-efficient Segmentation via Affinity Propagation". [NeurIPS2023]

7.【图像分割】On the Transferability of Learning Models for Semantic Segmentation for Remote Sensing Data

• 论文地址：https://arxiv.org//pdf/2310.10490

• 开源代码：GitHub - GDAOSU/Transferability-Remote-Sensing: This is the repo for paper "On the Transferability of Learning Models for Semantic Segmentation for Remote Sensing Images"

8.【目标跟踪】ZoomTrack: Target-aware Non-uniform Resizing for Efficient Visual Tracking

• 论文地址：https://arxiv.org//pdf/2310.10071

• 开源代码（即将开源）：GitHub - Kou-99/ZoomTrack: [NeurIPS 2023 Spotlight] ZoomTrack: Target-aware Non-uniform Resizing for Efficient Visual Tracking

9.【人脸识别】New Benchmarks for Asian Facial Recognition Tasks: Face Classification with Large Foundation Models

• 论文地址：https://arxiv.org//pdf/2310.09756

• 开源代码：GitHub - dukong1/KoIn_Benchmark_Dataset

10.【人脸识别】Learning Unified Representations for Multi-Resolution Face Recognition

• 论文地址：https://arxiv.org//pdf/2310.09563

• 开源代码：GitHub - StevenSmith2000/BTNet: [BMVC2023] The pytorch implementation for "Learning Unified Representations for Multi-Resolution Face Recognition"

11.【OCR】EfficientOCR: An Extensible, Open-Source Package for Efficiently Digitizing World Knowledge

• 论文地址：https://arxiv.org//pdf/2310.10050

• 工程主页：Efficient OCR

• 开源代码（即将开源）：https://github.com/dell-research-harvard/efficient_ocr

12.【点云】JM3D & JM3D-LLM: Elevating 3D Representation with Joint Multi-modal Cues

• 论文地址：https://arxiv.org//pdf/2310.09503

• 开源代码：GitHub - Mr-Neko/JM3D: The offical implemention of JM3D.

13.【医学图像处理】Generalizing Medical Image Representations via Quaternion Wavelet Networks

• 论文地址：https://arxiv.org//pdf/2310.10224

• 开源代码（即将开源）：GitHub - ispamm/QWT: Quaternion Wavelet Transform for Medical Imaging

14.【超分辨率重建】Image super-resolution via dynamic network

• 论文地址：https://arxiv.org//pdf/2310.10413

• 开源代码：GitHub - hellloxiaotian/DSRNet

15.【图像增强】Dimma: Semi-supervised Low Light Image Enhancement with Adaptive Dimming

• 论文地址：https://arxiv.org//pdf/2310.09633

• 开源代码：GitHub - WojciechKoz/Dimma: Dimma: Semi-supervised Low Light Image Enhancement with Adaptive Dimming

16.【场景文本识别】Scene Text Recognition Models Explainability Using Local Features

• 论文地址：https://arxiv.org//pdf/2310.09549

• 开源代码：GitHub - markytools/strexp

17.【动作识别】InfoGCN++: Learning Representation by Predicting the Future for Online Human Skeleton-based Action Recognition

• 论文地址：https://arxiv.org//pdf/2310.10547

• 开源代码：GitHub - stnoah1/infogcn2

18.【动作识别】3DYoga90: A Hierarchical Video Dataset for Yoga Pose Understanding

• 论文地址：https://arxiv.org//pdf/2310.10131

• 开源代码：GitHub - seonokkim/3DYoga90: Official repo of 3DYoga90 dataset

19.【领域泛化】Towards Unified and Effective Domain Generalization

• 论文地址：https://arxiv.org//pdf/2310.10008

• 工程主页：Towards Unified and Effective Domain Generalization

• 开源代码：GitHub - invictus717/UniDG: Towards Unified and Effective Domain Generalization

20.【多模态】Interpreting and Controlling Vision Foundation Models via Text Explanations

• 论文地址：https://arxiv.org//pdf/2310.10591

• 开源代码：GitHub - tonychenxyz/vit-interpret: Official implementation of "Interpreting and Controlling Vision Foundation Models via Text Explanations"

21.【多模态】Weakly Supervised Fine-grained Scene Graph Generation via Large Language Model

• 论文地址：https://arxiv.org//pdf/2310.10404

• 开源代码：GitHub - rlqja1107/torch-LLM4SGG: Utilizing Large Language Model for Weakly Supervised Scene Graph Generation

22.【多模态】LICO: Explainable Models with Language-Image Consistency

• 论文地址：https://arxiv.org//pdf/2310.09821

• 开源代码（即将开源）：GitHub - ymLeiFDU/LICO

23.【多模态】CAPro: Webly Supervised Learning with Cross-Modality Aligned Prototypes

• 论文地址：https://arxiv.org//pdf/2310.09761

• 开源代码：GitHub - yuleiqin/capro: This repo is the official released code of CAPro (NeurIPS 2023)

24.【多模态】LOVECon: Text-driven Training-Free Long Video Editing with ControlNet

• 论文地址：https://arxiv.org//pdf/2310.09711

• 开源代码：GitHub - zhijie-group/LOVECon: Official implementation for "LOVECon: Text-driven Training-free Long Video Editing with ControlNet"

25.【多模态】MiniGPT-v2: large language model as a unified interface for vision-language multi-task learning

• 论文地址：https://arxiv.org//pdf/2310.09478

• 工程主页：MiniGPT-v2

• 开源代码：GitHub - Vision-CAIR/MiniGPT-4: Open-sourced codes for MiniGPT-4 and MiniGPT-v2

26.【多模态】ViPE: Visualise Pretty-much Everything

• 论文地址：https://arxiv.org//pdf/2310.10543

• 开源代码（即将开源）：GitHub - Hazel1994/ViPE: SongAnimator

27.【多模态】RoboLLM: Robotic Vision Tasks Grounded on Multimodal Large Language Models

• 论文地址：https://arxiv.org//pdf/2310.10221

• 开源代码：GitHub - longkukuhi/armbench

28.【Diffusion】A Survey on Video Diffusion Models

• 论文地址：https://arxiv.org//pdf/2310.10647

• 开源代码：GitHub - ChenHsing/Awesome-Video-Diffusion-Models: [Arxiv] A Survey on Video Diffusion Models

29.【Diffusion】ConsistNet: Enforcing 3D Consistency for Multi-view Images Diffusion

• 论文地址：https://arxiv.org//pdf/2310.10343

• 工程主页：ConsistNet: Enforcing 3D Consistency for Multi-view Images Diffusion

• 开源代码（即将开源）：https://github.com/JiayuYANG/ConsistNet

30.【Diffusion】Scene Graph Conditioning in Latent Diffusion

• 论文地址：https://arxiv.org//pdf/2310.10338

• 开源代码：GitHub - FrankFundel/SGCond

31.【图像去噪】A cross Transformer for image denoising

• 论文地址：https://arxiv.org//pdf/2310.10408

• 开源代码：GitHub - hellloxiaotian/CTNet

32.【图像去噪】PUCA: Patch-Unshuffle and Channel Attention for Enhanced Self-Supervised Image Denoising

• 论文地址：https://arxiv.org//pdf/2310.10088

• 开源代码：GitHub - HyemiEsme/PUCA: hello

33.【人群计数】Semi-Supervised Crowd Counting with Contextual Modeling: Facilitating Holistic Understanding of Crowd Scenes

• 论文地址：https://arxiv.org//pdf/2310.10352

• 开源代码：GitHub - cha15yq/MRC-Crowd: Implementation of "Semi-Supervised Crowd Counting with Contextual Modeling: Facilitating Holistic Understanding of Crowd Scenes"

34.【NeRF】TraM-NeRF: Tracing Mirror and Near-Perfect Specular Reflections through Neural Radiance Fields

• 论文地址：https://arxiv.org//pdf/2310.10650

• 开源代码（即将开源）：GitHub - Rubikalubi/TraM-NeRF

35.【NeRF】DynVideo-E: Harnessing Dynamic NeRF for Large-Scale Motion- and View-Change Human-Centric Video Editing

• 论文地址：https://arxiv.org//pdf/2310.10624

• 工程主页：DynVideo-E: Harnessing Dynamic NeRF for Large-Scale Motion- and View-Change Human-Centric Video Editing

• 代码即将开源

36.【NeRF】ProteusNeRF: Fast Lightweight NeRF Editing using 3D-Aware Image Context

• 论文地址：https://arxiv.org//pdf/2310.09965

• 工程主页：ProteusNeRF

• 代码即将开源

37.【Continual Learning】Prior-Free Continual Learning with Unlabeled Data in the Wild

• 论文地址：https://arxiv.org//pdf/2310.10417

• 开源代码：GitHub - visiontao/pfcl

论文已打包，点击进入—>下载界面

CV计算机视觉交流群

群内包含目标检测、图像分割、目标跟踪、Transformer、多模态、NeRF、GAN、缺陷检测、显著目标检测、关键点检测、超分辨率重建、SLAM、人脸、OCR、生物医学图像、三维重建、姿态估计、自动驾驶感知、深度估计、视频理解、行为识别、图像去雾、图像去雨、图像修复、图像检索、车道线检测、点云目标检测、点云分割、图像压缩、运动预测、神经网络量化、网络部署等多个领域的大佬，不定期分享技术知识、面试技巧和内推招聘信息。

想进群的同学请添加微信号联系管理员：PingShanHai666。添加好友时请备注：学校/公司+研究方向+昵称。

推荐阅读：

CV计算机视觉每日开源代码Paper with code速览-2023.10.16

CV计算机视觉每日开源代码Paper with code速览-2023.10.13

港科大提出适用于夜间场景语义分割的无监督域自适应新方法

HSN：微调预训练ViT用于目标检测和语义分割，华南理工和阿里巴巴联合提出

EViT：借鉴鹰眼视觉结构，南开大学等提出ViT新骨干架构，在多个任务上涨点

如何优雅地读取网络的中间特征？