Git Product home page Git Product logo

Comments (5)

cmpute avatar cmpute commented on May 27, 2024 1

https://github.com/cmpute/d3d/blob/master/d3d/box/__init__.py
我写的版本,目前支持torch1.4+,理论上再往下支持也可以,只不过要改CMake

from rotated_iou.

lilanxiao avatar lilanxiao commented on May 27, 2024

你好,我只写了CUDA的版本。按照你的问题,我有三个建议,优先级依次递减:

  1. 尝试用低版本torch进行编译。我给torch的版本限定了1.5是因为我只在这个版本下测试过,低版本的torch可能也能编译。
  2. 尝试安装torch 1.5。无法安装torch 1.5常见的原因有Nvidia驱动版本不符,python环境冲突等等,经过排查一般是可以安装的。
  3. 如果实在搞不定可以自己用C++进行重构。因为CUDA版本和C++的核心逻辑是一致的, 只是CUDA版本可以自动并行化而已,在C下可以使用循环代替CUDA的并行运算。添加Pytorch C++拓展的具体操作可以参考这个教程。但是这是最不推荐的,因为不能使用GPU加速,数据在CPU和GPU之间来回传递也会产生Overhead。如果需要计算的Bounding Box多的话会非常慢。

祝好

from rotated_iou.

zzybj avatar zzybj commented on May 27, 2024

你好,我只写了CUDA的版本。按照你的问题,我有三个建议,优先级依次递减:

  1. 尝试用低版本torch进行编译。我给torch的版本限定了1.5是因为我只在这个版本下测试过,低版本的torch可能也能编译。
  2. 尝试安装torch 1.5。无法安装torch 1.5常见的原因有Nvidia驱动版本不符,python环境冲突等等,经过排查一般是可以安装的。
  3. 如果实在搞不定可以自己用C++进行重构。因为CUDA版本和C++的核心逻辑是一致的, 只是CUDA版本可以自动并行化而已,在C下可以使用循环代替CUDA的并行运算。添加Pytorch C++拓展的具体操作可以参考这个教程。但是这是最不推荐的,因为不能使用GPU加速,数据在CPU和GPU之间来回传递也会产生Overhead。如果需要计算的Bounding Box多的话会非常慢。

祝好

非常感谢您的回答,但是我用的是实验室的服务器,不敢自己改配置......,您感觉这个排序只用pytorch能实现吗? 我发现pytorch好像只有一个sort函数,无法实现自定义排序,而且最终还需要代码在模型运行时能够正常执行,不知道您有没有什么好的思路能分享一下,万分感谢

from rotated_iou.

zzybj avatar zzybj commented on May 27, 2024

你好,我只写了CUDA的版本。按照你的问题,我有三个建议,优先级依次递减:

  1. 尝试用低版本torch进行编译。我给torch的版本限定了1.5是因为我只在这个版本下测试过,低版本的torch可能也能编译。
  2. 尝试安装torch 1.5。无法安装torch 1.5常见的原因有Nvidia驱动版本不符,python环境冲突等等,经过排查一般是可以安装的。
  3. 如果实在搞不定可以自己用C++进行重构。因为CUDA版本和C++的核心逻辑是一致的, 只是CUDA版本可以自动并行化而已,在C下可以使用循环代替CUDA的并行运算。添加Pytorch C++拓展的具体操作可以参考这个教程。但是这是最不推荐的,因为不能使用GPU加速,数据在CPU和GPU之间来回传递也会产生Overhead。如果需要计算的Bounding Box多的话会非常慢。

祝好

大佬 您好 能帮忙看看这个错误吗 这是用torch1.5 ,install之后报的错误
running install
running bdist_egg
running egg_info
writing sort_vertices.egg-info/PKG-INFO
writing dependency_links to sort_vertices.egg-info/dependency_links.txt
writing top-level names to sort_vertices.egg-info/top_level.txt
reading manifest file 'sort_vertices.egg-info/SOURCES.txt'
writing manifest file 'sort_vertices.egg-info/SOURCES.txt'
installing library code to build/bdist.linux-x86_64/egg
running install_lib
running build_ext
building 'sort_vertices' extension
gcc -pthread -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -I/home/user1/anaconda3/lib/python3.6/site-packages/torch/include -I/home/user1/anaconda3/lib/python3.6/site-packages/torch/include/torch/csrc/api/include -I/home/user1/anaconda3/lib/python3.6/site-packages/torch/include/TH -I/home/user1/anaconda3/lib/python3.6/site-packages/torch/include/THC -I/usr/local/cuda-10.1/include -I/home/user1/anaconda3/include/python3.6m -c sort_vert.cpp -o build/temp.linux-x86_64-3.6/sort_vert.o -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=sort_vertices -D_GLIBCXX_USE_CXX11_ABI=0 -std=c++11
cc1plus: warning: command line option ‘-Wstrict-prototypes’ is valid for C/ObjC but not for C++
In file included from sort_vert.cpp:1:0:
utils.h:27:64: warning: backslash-newline at end of file
#define CHECK_IS_BOOL(x)
^
sort_vert.cpp: In function ‘at::Tensor sort_vertices(at::Tensor, at::Tensor, at::Tensor)’:
utils.h:12:73: error: ‘TORCH_CHECK’ was not declared in this scope
TORCH_CHECK(x.is_contiguous(), #x " must ne a contiguous tensor");
^
sort_vert.cpp:7:5: note: in expansion of macro ‘CHECK_CONTIGUOUS’
CHECK_CONTIGUOUS(vertices);
^
utils.h:12:73: error: ‘TORCH_CHECK’ was not declared in this scope
TORCH_CHECK(x.is_contiguous(), #x " must ne a contiguous tensor");
^
sort_vert.cpp:8:5: note: in expansion of macro ‘CHECK_CONTIGUOUS’
CHECK_CONTIGUOUS(mask);
^
utils.h:12:73: error: ‘TORCH_CHECK’ was not declared in this scope
TORCH_CHECK(x.is_contiguous(), #x " must ne a contiguous tensor");
^
sort_vert.cpp:9:5: note: in expansion of macro ‘CHECK_CONTIGUOUS’
CHECK_CONTIGUOUS(num_valid);
^
utils.h:7:61: error: ‘TORCH_CHECK’ was not declared in this scope
TORCH_CHECK(x.is_cuda(), #x " must be a CUDA tensor");
^
sort_vert.cpp:10:5: note: in expansion of macro ‘CHECK_CUDA’
CHECK_CUDA(vertices);
^
utils.h:7:61: error: ‘TORCH_CHECK’ was not declared in this scope
TORCH_CHECK(x.is_cuda(), #x " must be a CUDA tensor");
^
sort_vert.cpp:11:5: note: in expansion of macro ‘CHECK_CUDA’
CHECK_CUDA(mask);
^
utils.h:7:61: error: ‘TORCH_CHECK’ was not declared in this scope
TORCH_CHECK(x.is_cuda(), #x " must be a CUDA tensor");
^
sort_vert.cpp:12:5: note: in expansion of macro ‘CHECK_CUDA’
CHECK_CUDA(num_valid);
^
utils.h:24:49: error: ‘TORCH_CHECK’ was not declared in this scope
#x " must be a float tensor");
^
sort_vert.cpp:13:5: note: in expansion of macro ‘CHECK_IS_FLOAT’
CHECK_IS_FLOAT(vertices);
^
utils.h:30:48: error: ‘TORCH_CHECK’ was not declared in this scope
#x " must be a bool tensor");
^
sort_vert.cpp:14:5: note: in expansion of macro ‘CHECK_IS_BOOL’
CHECK_IS_BOOL(mask);
^
utils.h:18:47: error: ‘TORCH_CHECK’ was not declared in this scope
#x " must be a int tensor");
^
sort_vert.cpp:15:5: note: in expansion of macro ‘CHECK_IS_INT’
CHECK_IS_INT(num_valid);
^
sort_vert.cpp:23:54: error: expected primary-expression before ‘float’
sort_vertices_wrapper(b, n, m, vertices.data_ptr(), mask.data_ptr(),
^
sort_vert.cpp:23:78: error: expected primary-expression before ‘bool’
sort_vertices_wrapper(b, n, m, vertices.data_ptr(), mask.data_ptr(),
^
sort_vert.cpp:24:45: error: expected primary-expression before ‘int’
num_valid.data_ptr(), idx.data_ptr());
^
sort_vert.cpp:24:66: error: expected primary-expression before ‘int’
num_valid.data_ptr(), idx.data_ptr());
^
error: command 'gcc' failed with exit status 1

from rotated_iou.

lilanxiao avatar lilanxiao commented on May 27, 2024

关键的报错是这一句:

/bin/sh: /usr/local/cuda-8.0/bin/nvcc: No such file or directory

nvcc是nvidia的CUDA编译器,类似于c语言的gcc。这里的报错表示找不到nvcc。最可能的原因是没有安装cuda-toolkit。

安装方法参考官网的教程。注意必须使用10.2版本。

https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html

from rotated_iou.

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.