x-zhangyang / real-world-masked-face-dataset Goto Github PK
View Code? Open in Web Editor NEWReal-World Masked Face Dataset,口罩人脸数据集
Real-World Masked Face Dataset,口罩人脸数据集
链接: https://pan.baidu.com/s/1Vly3K-0qjlB6M2lenTZ8PA 提取码: xhze
您好,感谢整理口罩人脸数据集。
想问一下,能否告知关于口罩人脸数据集的使用许可和其他具体的使用条件呀?
I am only interested in Face Mask Detection not recognition, so Where can i find the MFDD Dataset mentioned in the research paper along with the labels for the bounding box of the mask?
The folders contain few non-mask images in the folder of with-mask. Please clean your dataset.
May I ask you, when will you open the source code of face recognition?
Hi. For non chinese people is really difficult to have a Baidu account.
So please could you upload to any other server??
Thanks,
Could you please share the link for the version of the CASIA_WebFace without Mask?
Because there are several versions for CASIA_WebFace. As far as I know, one version has only 455594 images, which is usually named 'Clean' version. The other two versions always are named raw_1 and raw_2 versions. For both of them, the number of images is 494414. Which version are you using to generate the simulated Mask database?
Could you please share your database for the original CASIA_WebFace(which has no mask)?
可否提供人脸的标注xml文件或类似于人脸标注的标注文件,看到提供的数据仅是图像
论文里提到的模型是如何训练的?
请问是否有提供统一的测试协议和测试脚本呢?LFW-Mask这类测试集,比对的是真是人脸和口罩人脸吗?
Hi, I can't download this link [https://pan.baidu.com/s/1P0PiWFNT1z_TcCj8vo43ow].
Could you please upload it to Google Drive?
agedb_30_masked的口罩数据集里面怎么还有没戴口罩的数据?这些没戴口罩的数据是原图吗?@X-zhangyang
In line 188 of wearmask.py, PIL's rotate function expects the rotation angle to be in degrees, but as np.arctan2 returns in units of radians, there is almost no rotation applied any time. This makes the code not work properly for tilted faces. Please look into it.
Thank you for offering the informative dataset for mask face recognition; it helps the developers like me save too much time collecting data.I've stared your project once I read the description just to show my support.
I have seen the demonstration video you put on GitHub, and it seems the accuracy of recognition is pretty high, so I wonder if I could ask for some details about the face detection and face recognition you used on this dataset, Maybe the name of the algorithm or framework, and if you've happened to have an open source project that implemented this function ,that would be great!
So, could you help me with that? anything helpful will be appreciated!
Why does the masked face dataset have 525 classes whereas the face dataset has 460 classes?
Hi, thanks for the contribution. May I know the license of this dataset? It would be great if you could add license in.
Firstly, thanks a lot for your amazing work.
But I feel some confusion about the dataset differences.
Could you please tell me what is the difference between dataset from link1 and dataset from link2?
Additionally, you have mentioned that for the Real-world masked face recognition dataset, the number of identities is 525. But I find that the number of identities in decompressed RWMFD(which is downloaded by cloning the GitHub repository) is 537. That is what makes me confused.
By the way, the Webface you used here refers to the CASIA-WebFace dataset, right?
Can I ask if the source code is open? I want to run this code, how?
webface lfw 模拟的戴口罩数据,戴的都是同一种口罩,这样数据训练的模型,会不会过分关注这种口罩,影响佩戴其他口罩的检测呢?您有验证过效果吗?
As we all know, the original LFW and CASIA-WebFace datasets have a dimension of 250*250.
Before carrying out the face recognition task, we should always perform face alignment on the original LFW and CACIA-WebFace dataset. For example, after face alignment, we obtained the alignment version datasets (the dimension of 112112) which have been aligned if I need to feed 112112 to neural network architecture.
I found that your Masked LFW and Masked CASIA-WebFace database have a dimension of 128*128.
Could you please explain more details about the detection and alignment you performed during the process of wearing the mask?
Do we still need to align again for masked face data? Or I do not need to perform face alignment again?
I downloaded Real-world masked face recognition dataset from https://drive.google.com/open?id=1UlOk6EtiaXTHylRUx2mySgvJX9ycoeBp.
However, I can't find the annotation labels of the face dataset. So I tried some regular face detection method such as MTCNN, dlib to dectect face and align face. But there are many wrong or missed detection probably due to the mask.
I wonder what kind of face detector you used to detect and align face with Real-world masked face recognition dataset. Thanks in advance.
Remarkable process in masked face recognition, but I note that may be not provide evaluation script, so how to evaluate this *_masked datasets? 1:1 face verification?
thanks
Hi,
I download this code with the help of one of my Chinese friend, but how to run this, also downloaded dlib dependencies, please tell me what arguments to pass and other dependencies.
Thank you for beautiful work
I'm looking to get the code of this project for my final year project and to run it on my computer, i'l be doing a few changes in order to increase the accuracy for it. Kindly reach me out at : [email protected]
您好,首先非常感谢您的工作!但是我发现真实口罩人脸数据集中有一些图像质量不高,例如有半张人脸的情况,请问这种情况是由于检测器导致的吗?可否将检测器检测到的区域适当外扩再做crop呢?
作者你好,很感谢你的工作。
由于github下载受限,是否能放一份原始数据到百度云网盘。
谢谢!
In the "11_0005_zhangjie_08" it seems that the unmasked picture is named as 15.jpg and in the pairs txt there is also a pair that contains 15.jpg and unmasked.jpg
@X-zhangyang 你好,可以分享下关于口罩人脸识别的模型和相关细节嘛?非常感谢~
Hello,想问一下95%的准确率是在哪个人脸识别的数据集测试的?LFW吗?是1:1的测试还是1:N的测试?
Hello!
Excuse me, but I can not find bounding boxes for detection dataset, could you please point on them.
你好,请问MFDD数据集的下载连接在哪里?
@X-zhangyang Would it be possible to re-upload at least the RMFD verification set to Google drive? Or somewhere?
I checked; it only contains 2203 masked faces and 90,468 normal faces.
Could you provide the labeled file?
As per the paper, "These datasets are freely available to industry and academia, based on which various applications on masked faces can be developed."
Could you please add a license file (MIT) or mention in readme about the license (CC0).
Thanks in advance for any response from you.
Thank you for sharing the link but Multiple Links do not work for Simulated masked face recognition datasets:
Could you please fix them to share new links to them?
Thank you
KMK
举几个例子,比如with mask部分的xiaoshenyang、liyifeng、chenweiting等等
如果这个数据作者还维护的话,希望可以清洗后重新发布。
给使用者提个醒,切勿直接拿来训练。
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.