Cheng Rui

Results 74 comments of Cheng Rui

> Okay, thanks for your reply, if I'm going to do a full-body gesture, I should also need an initial 3D coordinate similar to a Model .txt, do you know...

> Hi, in your arcface.txt file the model-engine-file field is arcface-mobilefacenet.engine, but I can't find that engine here. I want to ask you what model did you convert engine from?...

> Hi, I'm glad to see your update. As far as I know arcface's input is the aligned face. I would like to ask where did you implement the align...

> Hi, I'm glad to see your update. As far as I know arcface's input is the aligned face. I would like to ask where did you implement the align...

> Thank you for your reply. I am also looking for a way to do face alignment, if I succeed I will share! Great. Can you tell me a little...

你好。我在放假中,大概看了下你的问题,基本都是引用配置问题,问题一是yaml 的includes 没有配置好路径 是makefile或者cmake的路径问题;问题2是类似的OPENCV问题,你的opencv 没配置好。这个可能是我Makefile里少写了CV的引用,我误导你了 你看下centerface的CPP我记得解析解析没用到CV 你就别配置OPENCV了我后续会更新下 ------------------ 原始邮件 ------------------ 发件人: "positive666/Deepstream_Project" ***@***.***>; 发送时间: 2021年5月4日(星期二) 晚上8:09 ***@***.***>; ***@***.***>; 主题: [positive666/Deepstream_Project] 在Jetson Nano 上构建失败 (#2) 首先,十分感谢你的开源! 我想在Jetson Nano上复现你的项目,但出现了以下问题 这是我的环境 问题一:在构建RT模型时,cmake是成功的,当执行make时,报如下错误 这是什么原因呢? 问题二:在对nvdsinfer_custom_impl_CenterFace文件夹下的文件进行make操作时,MakeFile文件进行了如下修改 但make时一直报如下错误 但在/usr/include/opencv4/opencv2/路径下有opencv.hpp文件,这是什么原因呢? 最后,再次感谢你的分享,期待你的解答...

你的模型和Deepstream没关系,是Tensorrt你需要用tensorrt 去做转换,一个是ONNX到TRT的路线,一个是你把权重写进文件,然后自己写结构算子解析重构,这是tensortx有Demo参考。

想省事的话 用Yolov5的export去试试 可能需要调试修改下

Yes support, 4.0 release should require torch 1.7. The conversion model provides two project conversion methods under trans_project. Recently, I will supplement this project when I have time. You can...

之前没遇到过,但你说TensorRt没问题前提下,这里可能有两个具体原因?是否你的deepstream下的 解析cpp代码中NMS有问题 ;或者是阈值问题 你可以调一下看看,我在第一次作V3的模型转换时候遇到过,类似问题,当时是Nms的阈值造成的