seetaface / SeetaFaceEngine

Other
4.6k stars 1.78k forks source link

提取特征时如何利用GPU加速计算? #169

Open mender05 opened 7 years ago

mender05 commented 7 years ago

现在我测试得到的特征提取速度大概是500ms每张。项目中的identification模型好像是用caffe训练的。请问可以使用GPU加速特征提取吗?大概速度能提升多少呢?能否发布GPU版本的实现? 另,项目好像没人维护了,几乎一年没有更新过代码和添加新特性了。

HaapyDeng commented 6 years ago

同问,不知道有没有GPU版本的出来

ydar95 commented 6 years ago

我改写过 cuda版本 gtx780 50ms

HaapyDeng commented 6 years ago

你能分享发我一下你的代码吗?小弟感激不尽。

发送自 Windows 10 版邮件应用

发件人: 羊丹 发送时间: 2018年2月3日 11:52 收件人: seetaface/SeetaFaceEngine 抄送: HappyDeng; Comment 主题: Re: [seetaface/SeetaFaceEngine] 提取特征时如何利用GPU加速计算? (#169)

我改写过 cuda版本 gtx780 50ms — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.

HeshanMountteen commented 3 years ago

我改写过 cuda版本 gtx780 50ms

方便发我一份吗,最近在学习,希望能通过GPU优化

crystalssssssssssssssssssss commented 1 year ago

方便发我一份吗,最近在学习,希望能通过GPU优化