行人重识别库Open-ReID的下载和使用

博主自己也在github上提供了一个极好的行人重识别库,大家可以看看并下载和使用:

https://github.com/nickhuang1996/HJL-re-id

相关简介:行人重识别github开源库——HJL-re-id

详细内容参见Readme部分


目录

一、介绍

二、依赖项

三、下载

四、安装

五、使用前的注意事项

六、快速开始(训练+测试)


一、介绍

Open-ReID是一个轻量级行人重识别库,用于研究目的。

它旨在为不同的数据集提供统一的界面,一整套模型和评估指标,以及再现(接近)最新结果的示例。


二、依赖项

安装PyTorch(版本≥ 0.2.0)。

虽然Open-ReID同时支持python2和python3,但建议使用python3以获得更好的性能。


三、下载

1.下载地址为:https://cysu.github.io/open-reid/

2.我们下载它,并解压,如下:


四、安装

1.可以使用cmd运行:

cd open-reid
python setup.py install

2.使用IDE运行,可以debug得快一些,博主使用vs2019

关于如何使用vs2019运行Python程序,详情请看博主博客:vs2019 开始自己的第一个Python程序——九九乘法表

(1)脚本参数就是install

(2)运行结果

running install
running bdist_egg
running egg_info
writing open_reid.egg-info\PKG-INFO
writing dependency_links to open_reid.egg-info\dependency_links.txt
writing requirements to open_reid.egg-info\requires.txt
writing top-level names to open_reid.egg-info\top_level.txt
reading manifest file 'open_reid.egg-info\SOURCES.txt'
writing manifest file 'open_reid.egg-info\SOURCES.txt'
installing library code to build\bdist.win-amd64\egg
running install_lib
running build_py
creating build\bdist.win-amd64\egg
creating build\bdist.win-amd64\egg\reid
creating build\bdist.win-amd64\egg\reid\datasets
copying build\lib\reid\datasets\cuhk01.py -> build\bdist.win-amd64\egg\reid\datasets
copying build\lib\reid\datasets\cuhk03.py -> build\bdist.win-amd64\egg\reid\datasets
copying build\lib\reid\datasets\dukemtmc.py -> build\bdist.win-amd64\egg\reid\datasets
copying build\lib\reid\datasets\market1501.py -> build\bdist.win-amd64\egg\reid\datasets
copying build\lib\reid\datasets\viper.py -> build\bdist.win-amd64\egg\reid\datasets
copying build\lib\reid\datasets\__init__.py -> build\bdist.win-amd64\egg\reid\datasets
copying build\lib\reid\dist_metric.py -> build\bdist.win-amd64\egg\reid
creating build\bdist.win-amd64\egg\reid\evaluation_metrics
copying build\lib\reid\evaluation_metrics\classification.py -> build\bdist.win-amd64\egg\reid\evaluation_metrics
copying build\lib\reid\evaluation_metrics\ranking.py -> build\bdist.win-amd64\egg\reid\evaluation_metrics
copying build\lib\reid\evaluation_metrics\__init__.py -> build\bdist.win-amd64\egg\reid\evaluation_metrics
copying build\lib\reid\evaluators.py -> build\bdist.win-amd64\egg\reid
creating build\bdist.win-amd64\egg\reid\feature_extraction
copying build\lib\reid\feature_extraction\cnn.py -> build\bdist.win-amd64\egg\reid\feature_extraction
copying build\lib\reid\feature_extraction\database.py -> build\bdist.win-amd64\egg\reid\feature_extraction
copying build\lib\reid\feature_extraction\__init__.py -> build\bdist.win-amd64\egg\reid\feature_extraction
creating build\bdist.win-amd64\egg\reid\loss
copying build\lib\reid\loss\oim.py -> build\bdist.win-amd64\egg\reid\loss
copying build\lib\reid\loss\triplet.py -> build\bdist.win-amd64\egg\reid\loss
copying build\lib\reid\loss\__init__.py -> build\bdist.win-amd64\egg\reid\loss
creating build\bdist.win-amd64\egg\reid\metric_learning
copying build\lib\reid\metric_learning\euclidean.py -> build\bdist.win-amd64\egg\reid\metric_learning
copying build\lib\reid\metric_learning\kissme.py -> build\bdist.win-amd64\egg\reid\metric_learning
copying build\lib\reid\metric_learning\__init__.py -> build\bdist.win-amd64\egg\reid\metric_learning
creating build\bdist.win-amd64\egg\reid\models
copying build\lib\reid\models\inception.py -> build\bdist.win-amd64\egg\reid\models
copying build\lib\reid\models\resnet.py -> build\bdist.win-amd64\egg\reid\models
copying build\lib\reid\models\__init__.py -> build\bdist.win-amd64\egg\reid\models
copying build\lib\reid\trainers.py -> build\bdist.win-amd64\egg\reid
creating build\bdist.win-amd64\egg\reid\utils
creating build\bdist.win-amd64\egg\reid\utils\data
copying build\lib\reid\utils\data\dataset.py -> build\bdist.win-amd64\egg\reid\utils\data
copying build\lib\reid\utils\data\preprocessor.py -> build\bdist.win-amd64\egg\reid\utils\data
copying build\lib\reid\utils\data\sampler.py -> build\bdist.win-amd64\egg\reid\utils\data
copying build\lib\reid\utils\data\transforms.py -> build\bdist.win-amd64\egg\reid\utils\data
copying build\lib\reid\utils\data\__init__.py -> build\bdist.win-amd64\egg\reid\utils\data
copying build\lib\reid\utils\logging.py -> build\bdist.win-amd64\egg\reid\utils
copying build\lib\reid\utils\meters.py -> build\bdist.win-amd64\egg\reid\utils
copying build\lib\reid\utils\osutils.py -> build\bdist.win-amd64\egg\reid\utils
copying build\lib\reid\utils\serialization.py -> build\bdist.win-amd64\egg\reid\utils
copying build\lib\reid\utils\__init__.py -> build\bdist.win-amd64\egg\reid\utils
copying build\lib\reid\__init__.py -> build\bdist.win-amd64\egg\reid
byte-compiling build\bdist.win-amd64\egg\reid\datasets\cuhk01.py to cuhk01.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\datasets\cuhk03.py to cuhk03.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\datasets\dukemtmc.py to dukemtmc.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\datasets\market1501.py to market1501.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\datasets\viper.py to viper.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\datasets\__init__.py to __init__.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\dist_metric.py to dist_metric.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\evaluation_metrics\classification.py to classification.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\evaluation_metrics\ranking.py to ranking.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\evaluation_metrics\__init__.py to __init__.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\evaluators.py to evaluators.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\feature_extraction\cnn.py to cnn.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\feature_extraction\database.py to database.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\feature_extraction\__init__.py to __init__.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\loss\oim.py to oim.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\loss\triplet.py to triplet.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\loss\__init__.py to __init__.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\metric_learning\euclidean.py to euclidean.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\metric_learning\kissme.py to kissme.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\metric_learning\__init__.py to __init__.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\models\inception.py to inception.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\models\resnet.py to resnet.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\models\__init__.py to __init__.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\trainers.py to trainers.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\utils\data\dataset.py to dataset.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\utils\data\preprocessor.py to preprocessor.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\utils\data\sampler.py to sampler.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\utils\data\transforms.py to transforms.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\utils\data\__init__.py to __init__.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\utils\logging.py to logging.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\utils\meters.py to meters.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\utils\osutils.py to osutils.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\utils\serialization.py to serialization.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\utils\__init__.py to __init__.cpython-36.pyc
byte-compiling build\bdist.win-amd64\egg\reid\__init__.py to __init__.cpython-36.pyc
creating build\bdist.win-amd64\egg\EGG-INFO
copying open_reid.egg-info\PKG-INFO -> build\bdist.win-amd64\egg\EGG-INFO
copying open_reid.egg-info\SOURCES.txt -> build\bdist.win-amd64\egg\EGG-INFO
copying open_reid.egg-info\dependency_links.txt -> build\bdist.win-amd64\egg\EGG-INFO
copying open_reid.egg-info\requires.txt -> build\bdist.win-amd64\egg\EGG-INFO
copying open_reid.egg-info\top_level.txt -> build\bdist.win-amd64\egg\EGG-INFO
zip_safe flag not set; analyzing archive contents...
creating 'dist\open_reid-0.2.0-py3.6.egg' and adding 'build\bdist.win-amd64\egg' to it
removing 'build\bdist.win-amd64\egg' (and everything under it)
Processing open_reid-0.2.0-py3.6.egg
Removing d:\anaconda3\lib\site-packages\open_reid-0.2.0-py3.6.egg
Copying open_reid-0.2.0-py3.6.egg to d:\anaconda3\lib\site-packages
open-reid 0.2.0 is already the active version in easy-install.pth

Installed d:\anaconda3\lib\site-packages\open_reid-0.2.0-py3.6.egg
Processing dependencies for open-reid==0.2.0
Searching for metric-learn==0.4.0
Best match: metric-learn 0.4.0
Processing metric_learn-0.4.0-py3.6.egg
metric-learn 0.4.0 is already the active version in easy-install.pth

Using d:\anaconda3\lib\site-packages\metric_learn-0.4.0-py3.6.egg
Searching for scikit-learn==0.20.1
Best match: scikit-learn 0.20.1
Adding scikit-learn 0.20.1 to easy-install.pth file

Using d:\anaconda3\lib\site-packages
Searching for Pillow==5.3.0
Best match: Pillow 5.3.0
Adding Pillow 5.3.0 to easy-install.pth file

Using d:\anaconda3\lib\site-packages
Searching for h5py==2.8.0
Best match: h5py 2.8.0
Adding h5py 2.8.0 to easy-install.pth file

Using d:\anaconda3\lib\site-packages
Searching for six==1.11.0
Best match: six 1.11.0
Adding six 1.11.0 to easy-install.pth file

Using d:\anaconda3\lib\site-packages
Searching for torchvision==0.2.1
Best match: torchvision 0.2.1
Adding torchvision 0.2.1 to easy-install.pth file

Using d:\anaconda3\lib\site-packages
Searching for torch==1.0.1
Best match: torch 1.0.1
Adding torch 1.0.1 to easy-install.pth file
Installing convert-caffe2-to-onnx-script.py script to D:\Anaconda3\Scripts
Installing convert-caffe2-to-onnx.exe script to D:\Anaconda3\Scripts
Installing convert-onnx-to-caffe2-script.py script to D:\Anaconda3\Scripts
Installing convert-onnx-to-caffe2.exe script to D:\Anaconda3\Scripts

Using d:\anaconda3\lib\site-packages
Searching for scipy==1.1.0
Best match: scipy 1.1.0
Adding scipy 1.1.0 to easy-install.pth file

Using d:\anaconda3\lib\site-packages
Searching for numpy==1.15.4
Best match: numpy 1.15.4
Adding numpy 1.15.4 to easy-install.pth file

Using d:\anaconda3\lib\site-packages
Finished processing dependencies for open-reid==0.2.0

(3)查看项目文件夹,可以看到多了3个文件夹:

  • bulid
  • dist
  • open_reid.egg-info


五、使用前的注意事项

1.期间会下载VIPeR数据集,该数据集包含632个行人图像对,每个图像已缩放为128x48像素大小

Downloading http://users.soe.ucsc.edu/~manduchi/VIPeR.v1.0.zip to D:\vs2019_project\datasets\open-reid-master\examples\data\viper\raw\VIPeR.v1.0.zip

 分为2个文件夹:

  • cam_a
  • cam_b

打开这2个文件夹,可以看到cam_a主要拍摄的是行人的正面照cam_b主要拍摄是行人的侧身背身照 

(1)cam_a:

(2)cam_b:

2.此外,使用python3的用户需要更改reid文件夹底下的trainers.py文件的第33行,将

losses.update(loss.data[0], targets.size(0))

改为

losses.update(loss.data.item(), targets.size(0))

 如下


六、快速开始(训练+测试)

参数解释:

  • -d:dataset,设置为viper
  • -b:batch size,设置为64
  • -j:workers,设置为2(可根据cpu内核线程数而定)
  • -a:arch,设置为resnet50(模型名称,可选'inception', 'resnet18', 'resnet34', 'resnet50', 'resnet101',和 'resnet152')

1.cmd运行examples文件夹底下的softmax_loss.py文件

python examples/softmax_loss.py -d viper -b 64 -j 2 -a resnet50 --logs-dir logs/softmax-loss/viper-resnet50

2.IDE运行,vs2019中设置softmax_loss.py文件为启动文件

3.运行程序:

训练,默认50个epoch

最后进行测试

4.此外,还会生成logs文件夹,如下:

5.我们的实验记录在logs\softmax-loss\viper-resnet50可以看到:

  • 有每次训练的checkpoint
  • 最好的checkpoint
  • 实验记录log.txt(记录了控制台打印的实验信息)


是不是很好用呢~

悲恋花丶无心之人 CSDN认证博客专家 深度学习 神经网络 Pytorch
计算机视觉在读研究生,熟悉Pytorch,MXNet,TensorFlow,Keras等深度学习框架,主要涉及的领域有目标检测,语义分割,超分辨率重建,行人重识别等。
个人GitHub网址为:https://github.com/nickhuang1996
©️2020 CSDN 皮肤主题: Age of Ai 设计师:meimeiellie 返回首页
实付 29.90元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、C币套餐、付费专栏及课程。

余额充值