OCR——视觉会议调研_icpr会议级别

作者：空白诗007 | 2024-07-23 17:22:13

踩

icpr会议级别

视觉会议：

（以下只详细描述ocr相关的竞赛）

国际计算机视觉大会（ICCV），计算机视觉模式识别会议（CVPR）和欧洲计算机视觉会议（ECCV）并称计算机视觉方向的三大顶级会议。

ICCV是公认的三个会议中级别最高的。不同于CVPR举办地没有出过美国、ECCV仅限欧洲，它的举办地点会在世界各地选择，赞助也不少。

ICPR（模式识别领域的顶级国际会议，CCF C类会议，两年一届）

由国际权威学术组织---国际模式识别协会(The International Association for Pattern Recognition，IAPR)主办

ICPR2020（共10个 Challenges，意大利米兰，截稿日2020-07-15，会议日2021-01-10）：

Challenge: 3
AcTiVComp20
Competition on Superimposed Text Detection and Recognition in Arabic News Video Frames 阿拉伯新闻视频帧中叠加文本检测和识别的竞赛

task1 阿拉伯新闻视频帧中的文本检测简介：

到目前为止，ocr方法仅针对某些语言，例如拉丁语和中文。对于阿拉伯语这样的语言，全球有十亿多人在使用它，文献仅限于很少的研究，并且大多数现有方法已在具有非统一评估协议的私有数据集上进行了测试，这使得直接比较和科学基准测试不切实际。本竞赛旨在通过鼓励Scene / Video-OCR研究人员在标准的带注释数据集上开发并测试其系统。（AcTiV数据集，行级别标注）

该任务的目的是根据边界框<x，y，w，h>获得视频帧中文本区域的估计。

AcTiV-D数据集：

注：没有用于滚动文本栏的groundtruth信息

标注：x，y坐标（左上像素），宽度和高度

task2 裁剪新闻图像中的文本识别简介：

以文本行图像作为输入，此任务的目的是生成相应的文本转录。

AcTiV-R数据集中的一些文本行图像示例：

标签：

XML文件由两个主要的标记部分组成：ArabicTranscription和LatinTranscription。为了使阿拉伯语文本易于访问，将其转换为带有后缀的一组拉丁语标签。这些后缀表示字母在单词中的位置，即B：开始，M：中间，E：结束，I：隔离。这两个阿拉伯语单词“نشرةالأخبار”的拉丁语翻译为“ Nuun_B Shiin_M Raa_E TaaaClosed_I Space Alif_I Laam_EHamzaAboveAlif_E Xaa_B Baa_M Alif_E Raa_I”。

task3 端到端阿拉伯视频文本识别简介：

AcTiV数据集（标注格式与task1的有所不同）

官网：

https://www.icpr2020.it/

https://diuf.unifr.ch/main/diva/AcTiVComp/evaluation.html

ICPR2018（北京，截稿日--，会议日2018-08-20持续一周）：

官网：

http://www.icpr2018.org/ 已关闭

CVPR（IEEE Conference on Computer Vision and Pattern Recognition 计算机视觉和模式识别大会，CCF A类会议，一年一届）

CVPR2021（主题包括计算机视觉和模式识别的所有方面，Nashville, Tennessee, USA，截稿日2020-11-16，会议日2021-06-10—2021-06-25）：

包括但不限于：

2D object recognition
3D computer vision
3D object recognition
Action and behavior recognition
Adversarial learning, adversarial attack and defense methods
Biometrics, face, gesture, body pose
Computational photography
Datasets and evaluation
Efficient training and inference methods for networks
Explainable AI, fairness, accountability, privacy, transparency and ethics in vision
Image and video retrieval
Image and video synthesis
Image classification
Low-level and physics-based vision
Machine learning architectures and formulations
Medical, biological and cell microscopy
Motion and tracking
Optimization and learning methods
Pose estimation
Representation learning, deep learning
Scene analysis and understanding
Transfer, low-shot, semi- and un- supervised learning
Video analysis and understanding
Vision + language, vision + other modalities
Vision applications and systems, vision for robotics and autonomous vehicles

官网：

http://cvpr2021.thecvf.com/node/33

CVPR2019 （加利福尼亚州，截稿日2018-11-16，会议日2019-6-18-2019-6-20）

官网：

https://cvpr2019.thecvf.com/submission/timeline

领域：无展示

ECCV（European Conference on Computer Vision 欧洲计算机视觉会议，CCF B类会议，两年一届）

ECCV2020（格拉斯哥，截稿日2020-03-05，会议日2020-08-23）：

包括

3D Poses In the Wild Challenge
4D Vision
Adversarial Robustness in the Real World
BioImage Computation
Bodily Expressed Emotion Understanding
Commands 4 Autonomous Vehicles
Computer VISion for ART Analysis
Egocentric Perception, Interaction and Computing
Embodied Vision, Actions & Language
Eye Gaze in VR, AR, and in the Wild
Holistic Scene Structures for 3D Vision
International Challenge on Compositional and Multimodal Perception
Joint COCO and LVIS Recognition Challenge
Multimodal Video Analysis Workshop and Moments in Time Challenge
Object Tracking and its Many Guises
Perception for Autonomous Driving
Recovering 6D Object Pose
SHApe Recovery from Partial Textured 3D Scans
Sign Language Recognition, Translation and Production
TASK-CV Workshop and VisDA Challenge
Visual Inductive Priors for Data-Efficient Deep Learning
Women in Computer Vision
Learning": Training with no prior examples
Advances in Image Manipulation Workshop and Challenges
Assistive Computer Vision and Robotics
Benchmarking Trajectory Forecasting Models
Beyond mAP: Reasessing the Evaluation of Object Detection
Computer Vision for UAVs Workshop and Challenge
Computer Vision Problems in Plant Phenotyping
Embedded Vision
Fair Face Recognition and Analysis
GigaVision: When Gigapixel Videography Meets Computer Vision
Imbalance Problems in Computer Vision
Instance-Level Recognition
Learning 3D Representations for Shape and Appearance
Long-Term Visual Localization under Changing Conditions
Perception Through Structured Generative Models
Robust Vision Challenge 2020
Sensing, Understanding and Synthesizing Humans
The Bright and Dark Sides of Computer Vision: Challenges and Opportunities for Privacy and Security
The Visual Object Tracking Challenge
Video Turing Test: Toward Human-Level Video Story Understanding