Coder Social home page Coder Social logo

在standford cars 数据集上,测试正确率88.6%左右就上不去了,有达到论文中的93.9%的吗,batch size为8 about nts-net HOT 20 OPEN

dongzhi0312 avatar dongzhi0312 commented on June 19, 2024
在standford cars 数据集上,测试正确率88.6%左右就上不去了,有达到论文中的93.9%的吗,batch size为8

from nts-net.

Comments (20)

heartInsert avatar heartInsert commented on June 19, 2024

我内存爆炸了我去,你运行环境多大内存的

from nts-net.

novioleo avatar novioleo commented on June 19, 2024

@heartInsert

我16G内存,16Gswap能撑住cub200

from nts-net.

dongzhi0312 avatar dongzhi0312 commented on June 19, 2024

12G 内存,batch size 为 12

from nts-net.

heartInsert avatar heartInsert commented on June 19, 2024

我改了dataset,只有在getittem的时候才读取Img,而不是一开始就读取img

from nts-net.

novioleo avatar novioleo commented on June 19, 2024

@heartInsert 如果能做预加载就很好了

from nts-net.

heartInsert avatar heartInsert commented on June 19, 2024

@heartInsert 如果能做预加载就很好了
啥?

from nts-net.

novioleo avatar novioleo commented on June 19, 2024

@heartInsert 你每次get_item再去读取图片,会从某个角度影响性能,决定于磁盘io。如果你多个worker还行,如果不是的话,就需要预加载,就是让后台去提前加载这部分图像到内存,然后到时内存直接读取,而不需要等待了。

from nts-net.

heartInsert avatar heartInsert commented on June 19, 2024

@heartInsert 你每次get_item再去读取图片,会从某个角度影响性能,决定于磁盘io。如果你多个worker还行,如果不是的话,就需要预加载,就是让后台去提前加载这部分图像到内存,然后到时内存直接读取,而不需要等待了。

00,对的,这样确实比较慢,直接读内存那种肯定快好多

from nts-net.

lyn0102 avatar lyn0102 commented on June 19, 2024

@heartInsert 你每次get_item再去读取图片,会从某个角度影响性能,决定于磁盘io。如果你多个worker还行,如果不是的话,就需要预加载,就是让后台去提前加载这部分图像到内存,然后到时内存直接读取,而不需要等待了。

00,对的,这样确实比较慢,直接读内存那种肯定快好多

你有遇到训练和测试在测试集上的结果不一样的情况么?

from nts-net.

novioleo avatar novioleo commented on June 19, 2024

@lyn0102 能具体说下么?或者po图。

from nts-net.

lyn0102 avatar lyn0102 commented on June 19, 2024

@lyn0102 能具体说下么?或者po图。

应该是他的测试代码有问题,少了个eval

from nts-net.

lunchben avatar lunchben commented on June 19, 2024

No description provided.

您好,请问怎么在Stanford cars上训练啊,能分享一下您的dataset文件吗?

from nts-net.

yangxh11 avatar yangxh11 commented on June 19, 2024

请问问题解决了吗,我也遇到这个问题,stanford cars的测试效果上不去,不管是否使用该数据集提供的bbox标签

from nts-net.

haobabuhaoba avatar haobabuhaoba commented on June 19, 2024

No description provided.

您好,请问怎么在Stanford cars上训练啊,能分享一下您的dataset文件吗?

请问,您现在会修改成自己的数据集了吗?

from nts-net.

lunchben avatar lunchben commented on June 19, 2024

No description provided.

您好,请问怎么在Stanford cars上训练啊,能分享一下您的dataset文件吗?

请问,您现在会修改成自己的数据集了吗?

我学了pytorch之后懒得写dataloader,把自己的数据集用python脚本写成CUB200的格式然后直接用源代码运行就行了。

from nts-net.

haobabuhaoba avatar haobabuhaoba commented on June 19, 2024

from nts-net.

lunchben avatar lunchben commented on June 19, 2024

不太会,你能发给我一个脚本吗?我试着把自己的数据集改成CUB200的格式的,我的是VOC格式的数据集

------------------ 原始邮件 ------------------ 发件人: "yangze0930/NTS-Net" <[email protected]>; 发送时间: 2020年11月13日(星期五) 上午9:53 收件人: "yangze0930/NTS-Net"<[email protected]>; 抄送: "张冰华"<[email protected]>;"Comment"<[email protected]>; 主题: Re: [yangze0930/NTS-Net] 在standford cars 数据集上,测试正确率88.6%左右就上不去了,有达到论文中的93.9%的吗,batch size为8 (#17) No description provided. 您好,请问怎么在Stanford cars上训练啊,能分享一下您的dataset文件吗? 请问,您现在会修改成自己的数据集了吗? 我学了pytorch之后懒得写dataloader,把自己的数据集用python脚本写成CUB200的格式然后直接用源代码运行就行了。 — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

链接:https://pan.baidu.com/s/10Wc1KmtlFDVCgFQFnNq6-A
提取码:Lv19

from nts-net.

Opalus-wangs avatar Opalus-wangs commented on June 19, 2024

No description provided.

您好,请问怎么在Stanford cars上训练啊,能分享一下您的dataset文件吗?

请问,您现在会修改成自己的数据集了吗?

我学了pytorch之后懒得写dataloader,把自己的数据集用python脚本写成CUB200的格式然后直接用源代码运行就行了。

您好,我想请问一下,我想换成Standford Dog数据集,仿照CUB生成了那三个.txt(images.txt...)文件,思路是可以的么?但文件读取一直不合适,能分享一下您的脚本文件嘛?万分感谢

from nts-net.

lunchben avatar lunchben commented on June 19, 2024

No description provided.

您好,请问怎么在Stanford cars上训练啊,能分享一下您的dataset文件吗?

请问,您现在会修改成自己的数据集了吗?

我学了pytorch之后懒得写dataloader,把自己的数据集用python脚本写成CUB200的格式然后直接用源代码运行就行了。

您好,我想请问一下,我想换成Standford Dog数据集,仿照CUB生成了那三个.txt(images.txt...)文件,思路是可以的么?但文件读取一直不合适,能分享一下您的脚本文件嘛?万分感谢

链接: https://pan.baidu.com/s/1F3nuKOod05nH5K7VYtPrIg 提取码: bd98
脚本有点low,你要根据自己的路径和文件名改一下

from nts-net.

durong1996 avatar durong1996 commented on June 19, 2024

No description provided.

您好,请问怎么在Stanford cars上训练啊,能分享一下您的dataset文件吗?

请问,您现在会修改成自己的数据集了吗?

我学了pytorch之后懒得写dataloader,把自己的数据集用python脚本写成CUB200的格式然后直接用源代码运行就行了。

您好,我想请问一下,我想换成Standford Dog数据集,仿照CUB生成了那三个.txt(images.txt...)文件,思路是可以的么?但文件读取一直不合适,能分享一下您的脚本文件嘛?万分感谢

链接: https://pan.baidu.com/s/1F3nuKOod05nH5K7VYtPrIg 提取码: bd98 脚本有点low,你要根据自己的路径和文件名改一下

可以麻烦再发一下您数据集的链接吗?就是Standford Dog数据集和Stanford cars数据集,特别感谢您

from nts-net.

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.