Giter Site home page Giter Site logo

Comments (49)

github-luffy avatar github-luffy commented on June 3, 2024

不好意思 这个模型是不能用的 我重新上传一个模型文件 你试试看

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

上传的模型暂时只是用了300W与WFLW训练出来的，68点特征点 你试试看 best_model.pth.

from pfld_68points_pytorch.

DWCTOD avatar DWCTOD commented on June 3, 2024

感谢大佬回复和分析,我想试着按照项目中的介绍去训练一下,发现在数据集处理的时候就出现问题了
image
把下载好的数据集放在这个路径下然后运行 python3 WFLW_SetPreparation68.py 报错,根据提示应该是路径的问题导致图片没有读取进去

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

感谢大佬回复和分析,我想试着按照项目中的介绍去训练一下,发现在数据集处理的时候就出现问题了
image
把下载好的数据集放在这个路径下然后运行 python3 WFLW_SetPreparation68.py 报错,根据提示应该是路径的问题导致图片没有读取进去

图片是放在/data/WFLW/raw里面嘛 raw里面放的应该是WFLW的图片目录 是这样的嘛?raw目录下存放的目录应该是如下所示：
0--Parade
1--Handshaking
...

from pfld_68points_pytorch.

DWCTOD avatar DWCTOD commented on June 3, 2024

感谢大佬回复和分析,我想试着按照项目中的介绍去训练一下,发现在数据集处理的时候就出现问题了
image
把下载好的数据集放在这个路径下然后运行 python3 WFLW_SetPreparation68.py 报错,根据提示应该是路径的问题导致图片没有读取进去

图片是放在/data/WFLW/raw里面嘛 raw里面放的应该是WFLW的图片目录 是这样的嘛?

O.O,原来是这样子的,哈哈哈,因为我解压之后还有一个大的文件夹包含里面的小文件,现在可以啦哈,太感谢啦

from pfld_68points_pytorch.

DWCTOD avatar DWCTOD commented on June 3, 2024

不知道大佬对于关键点的抖动,有没有什么好的思路去解决呢? 加跟踪?

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

关键点的抖动 是指 测试视频流中的人脸 其特征点抖动的比较厉害？

from pfld_68points_pytorch.

DWCTOD avatar DWCTOD commented on June 3, 2024

关键点的抖动 是指 测试视频流中的人脸 其特征点抖动的比较厉害？

是的呀,就是人基本不怎么动的情况下,人脸特征点还是会一直抖动的比较厉害

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

刚上传的模型 你测试也会这样的嘛

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

1.要保证人脸框不能抖得太厉害，就是前后人脸框位置变化大，当然真实人脸位置移动正常，可以加人脸跟踪方法，人脸跟踪其实也可简单设置；
2.可能人脸特征点模型泛化能力不强，可增加数据训练。

from pfld_68points_pytorch.

DWCTOD avatar DWCTOD commented on June 3, 2024

是呀,其实这种抖动,就是人基本正常不怎么动的时候也会存在的

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

你好 @github-luffy ，我想问一下我根据已有98点（https://github.com/polarisZhao/PFLD-pytorch）代码训练出来的98点的NME只有0.07左右，我看到你在另外一个工程里说到的调优方法，不知道对应的具体的网络是哪个？

谢谢

mobileNetV3_large或者blazeface,你都可以试试

from pfld_68points_pytorch.

zcxxlshirley avatar zcxxlshirley commented on June 3, 2024

请问
运行 camera.py
抛出terminate called after throwing an instance of 'dmlc::Error'
Aborted (core dumped)
是怎么回事呢？

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

请问
运行 camera.py
抛出terminate called after throwing an instance of 'dmlc::Error'
Aborted (core dumped)
是怎么回事呢？

用了mxnet，你可能安装的不是gpu版本，先卸载mxnet，然后安装mxnet-gpu版本

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

EfficientNet系列的模型已更新，速度快，精度高，奥力给！

from pfld_68points_pytorch.

lfxx avatar lfxx commented on June 3, 2024

EfficientNet系列的模型已更新，速度快，精度高，奥力给！

您好，更新后的预训练模型有吗，想测试一下效果．测试了best_model.pth,感觉效果不是很好.另外想问您一下，输出的关键点置信度有什么意义呢，表示点位置的精准度吗．使用best_model.pth测试时关键点置信度并不是很高

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

best_model.pth是利用300W与WFLW两个训练集训练出来的最好模型，使用BlazeLandmark网络，应该还可以继续调优。
输出的关键点置信度是指关键点热力图，正常来说置信度越高，表示是人脸关键点的可能性越高。

from pfld_68points_pytorch.

lfxx avatar lfxx commented on June 3, 2024

best_model.pth是利用300W与WFLW两个训练集训练出来的最好模型，使用BlazeLandmark网络，应该还可以继续调优。
输出的关键点置信度是指关键点热力图，正常来说置信度越高，表示是人脸关键点的可能性越高。

谢谢您的回复，看了下源码，这个置信度其实并不是图像分类中的置信度，其实是位置归一化后的数值，乘上宽高以后就可以得到点位的坐标．这样理解对吗？
另外如果训练时的人脸提取器和推理时的不一样，比如用mtcnn提取人脸位置用于训练,retinaface提取人脸用于推理，会影响检测效果吗？

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

训练特征点时需要归一化操作，编码与解码的一个过程。
训练样本不一样，检测效果肯定不一样。

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

image
能把这个文件发一下吗？运行camera.py就出现这个错。

安装了mxnet-gpu吗

from pfld_68points_pytorch.

17601614169 avatar 17601614169 commented on June 3, 2024

from pfld_68points_pytorch.

17601614169 avatar 17601614169 commented on June 3, 2024

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

python3.6

from pfld_68points_pytorch.

17601614169 avatar 17601614169 commented on June 3, 2024

from pfld_68points_pytorch.

17601614169 avatar 17601614169 commented on June 3, 2024

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

博主您好，如果我想用图片测试的话需要运行哪个代码？
…
------------------ 原始邮件 ------------------ 发件人: "github-luffy/PFLD_68points_Pytorch" <[email protected]>; 发送时间: 2020年8月27日(星期四) 下午2:51 收件人: "github-luffy/PFLD_68points_Pytorch"<[email protected]>; 抄送: "♚丨勿忘心安💢"<[email protected]>;"Comment"<[email protected]>; 主题: Re: [github-luffy/PFLD_68points_Pytorch] camera.py运行自动闪退了 (#1) 能把这个文件发一下吗？运行camera.py就出现这个错。 安装了mxnet-gpu吗 — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

test.py,代码估计也好懂 就不讲了

from pfld_68points_pytorch.

17601614169 avatar 17601614169 commented on June 3, 2024

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

好的，谢谢。

test.py,代码估计也好懂 就不讲了

from pfld_68points_pytorch.

17601614169 avatar 17601614169 commented on June 3, 2024

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

辛苦了。
…
------------------ 原始邮件 ------------------ 发件人: "github-luffy/PFLD_68points_Pytorch" <[email protected]>; 发送时间: 2020年8月27日(星期四) 下午5:09 收件人: "github-luffy/PFLD_68points_Pytorch"<[email protected]>; 抄送: "♚丨勿忘心安💢"<[email protected]>;"Comment"<[email protected]>; 主题: Re: [github-luffy/PFLD_68points_Pytorch] camera.py运行自动闪退了 (#1) 好的，谢谢。 test.py,代码估计也好懂 就不讲了 — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

不辛苦 点个赞就ok 哈

from pfld_68points_pytorch.

Shzhentao avatar Shzhentao commented on June 3, 2024

您好，我运行python camera.py 显示No such file or directory: 'test_models2/model_37.pth'，这是什么问题呢？谢谢！

from pfld_68points_pytorch.

17601614169 avatar 17601614169 commented on June 3, 2024

from pfld_68points_pytorch.

Shzhentao avatar Shzhentao commented on June 3, 2024

没有呀

from pfld_68points_pytorch.

17601614169 avatar 17601614169 commented on June 3, 2024

from pfld_68points_pytorch.

Shzhentao avatar Shzhentao commented on June 3, 2024

你成功了吗？它显示runtime error: ./pretrained_model/efficientNet-b0.pth is a zip archive，你是加载了哪个

from pfld_68points_pytorch.

17601614169 avatar 17601614169 commented on June 3, 2024

image

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

您好，我运行python camera.py 显示No such file or directory: 'test_models2/model_37.pth'，这是什么问题呢？谢谢！

你可以试试 pretrained_model的几个模型，记得加载模型
image

from pfld_68points_pytorch.

17601614169 avatar 17601614169 commented on June 3, 2024

我没训练直接跑的test代码，但没成功，根据网上的方法都试过了也不行。
image

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

我没训练直接跑的test代码，但没成功，根据网上的方法都试过了也不行。
image

人脸检测的retinaface模型是mxnet框架的，你安装的是哪个版本的mxnet，mxnet还是mxnet-gpu？

from pfld_68points_pytorch.

17601614169 avatar 17601614169 commented on June 3, 2024

我没训练直接跑的test代码，但没成功，根据网上的方法都试过了也不行。
图片

人脸检测的retinaface模型是mxnet框架的，你安装的是哪个版本的mxnet，mxnet还是mxnet-gpu？

image
cpu

from pfld_68points_pytorch.

17601614169 avatar 17601614169 commented on June 3, 2024

电脑不是N卡 装gpu也没用吧？

from pfld_68points_pytorch.

Shzhentao avatar Shzhentao commented on June 3, 2024

您好，请问您用的pytorch版本是多少呀？

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

试试安装一下mxnet-gpu,然后跑跑看

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

pytorch 1.0以上

from pfld_68points_pytorch.

Shzhentao avatar Shzhentao commented on June 3, 2024

谢谢您，但是我这里加载模型会显示RuntimeError: ./pretrained_model/blazelandmark.pth is a zip archive(did you mean to use torch.jit.load())
question

from pfld_68points_pytorch.

Shzhentao avatar Shzhentao commented on June 3, 2024

好像说是模型是zip archive架构的

from pfld_68points_pytorch.

github-luffy avatar github-luffy commented on June 3, 2024

干脆直接用最新的pytorch版本好了

from pfld_68points_pytorch.

Shzhentao avatar Shzhentao commented on June 3, 2024

好的，谢谢您，我试试

from pfld_68points_pytorch.

17601614169 avatar 17601614169 commented on June 3, 2024

image
换成gpu就出这个错了，头大

from pfld_68points_pytorch.

Related Issues (20)

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.