Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

关于UpdateNet训练连数据集问题 #24

Closed
ShiLongCheng opened this issue Dec 12, 2020 · 8 comments
Closed

关于UpdateNet训练连数据集问题 #24

ShiLongCheng opened this issue Dec 12, 2020 · 8 comments

Comments

@ShiLongCheng
Copy link

您好,我用您提供的程序在ubuntu16.04下面生成训练数据时出现MemoryError问题。当我按照网上的说明 cat /proc/sys/vm/overcommit_memory改为1时,又显示已杀死,没法生成训练数据。请问您是否遇到过同样的问题。或者您是否能够把训练数据分享一下,谢谢。

@ShiLongCheng
Copy link
Author

您好,这个问题我又解决了,是内存不够的问题,还是得加硬件才行。借了实验室小伙伴的内存条凑成了32g内存。据我实际操作,按照您提供的那20个视频数据生成训练数据时,大概要用23g左右的运行内存。

@HonglinChu
Copy link
Owner

HonglinChu commented Dec 14, 2020 via email

@HonglinChu
Copy link
Owner

HonglinChu commented Dec 14, 2020 via email

@ShiLongCheng
Copy link
Author

我没遇到过这个问题,你用vscode单步调试一下,找找是那一步代码错了

------------------ 原始邮件 ------------------ 发件人: "HonglinChu/SiamTrackers" <notifications@github.com>; 发送时间: 2020年12月12日(星期六) 晚上11:50 收件人: "HonglinChu/SiamTrackers"<SiamTrackers@noreply.github.com>; 抄送: "Subscribed"<subscribed@noreply.github.com>; 主题: [HonglinChu/SiamTrackers] 关于UpdateNet训练连数据集问题 (#24) 您好,我用您提供的程序在ubuntu16.04下面生成训练数据时出现MemoryError问题。当我按照网上的说明 cat /proc/sys/vm/overcommit_memory改为1时,又显示已杀死,没法生成训练数据。请问您是否遇到过同样的问题。或者您是否能够把训练数据分享一下,谢谢。 — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

您好,这个问题我已经解决了。但是我生成训练数据后,训练时始终没有达到0.403,最多0.363好像。不知道为什么。是我哪里没考虑到吗?还有您哪个README里面讲的我没有看太懂。每个阶段怎么训练的?

@ShiLongCheng
Copy link
Author

ShiLongCheng commented Dec 14, 2020

还有一点,我是觉得updatenet只用了两层1x1的效果就挺好的了,感觉不太合理啊。毕竟两层线性的网络,而且是1x1的网络融合能力有点差吧,原作者为什么不加深一下? 趁着您在线,多问几个问题。^_^

@ShiLongCheng ShiLongCheng reopened this Dec 14, 2020
@HonglinChu
Copy link
Owner

HonglinChu commented Dec 15, 2020 via email

@ShiLongCheng
Copy link
Author

我没尝试过加深,应该是两层就够了,两层1x1,只用了一个relu,个人感觉非线性表达不够。Updatenet主要是进行通道交互,1x1卷积进行通道降维和升维来实现。关于updatenet可以微信上交流

------------------ 原始邮件 ------------------ 发件人: "alfredcheng"<notifications@github.com>; 发送时间: 2020年12月14日(星期一) 下午3:10 收件人: "HonglinChu/SiamTrackers"<SiamTrackers@noreply.github.com>; 抄送: "SmallLion🦁️"<1628464345@qq.com>; "Comment"<comment@noreply.github.com>; 主题: Re: [HonglinChu/SiamTrackers] 关于UpdateNet训练连数据集问题 (#24) 还有一点,我是觉得updatenet只用了两层1x1的效果就挺好的了,感觉不太合理啊。毕竟两层线性的网络,而且是1x1的网络融合能力有点差吧,原作者为什么不加深一下? — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

您好,您的微信号是多少啊,我没找到。^-^

@HonglinChu
Copy link
Owner

HonglinChu commented Dec 15, 2020 via email

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants