Java-K8s-Quarkus
1 | #安装 ---废弃开始----直接可以在idea里面新建quarkus的项目 |
0%
记一次从vmdk镜像里面恢复数据的历程
问题描述
- 环境： ESXi 7.0+centos7虚拟机
在一次操作生产服务器时，发生了系统文件丢失，导致重启之后无法进入系统，但是重要哦的文件还在系统中，第一时间应该先对当前系统【创建快照】，创建快照之后就可以【查看vmdk文件】
解决方案一：导出vmdk文件进行打开（废弃）
- 首先需要导出vmdk文件，可以【开启ESXI服务器的ssh功能】，然后使用
scp root@<ESXi_IP>:/vmfs/volumes/datastore_name/vm_name/disk.vmdk /local/path/拷贝出来，如果文件太大，可以【创建虚拟机】作为临时存放，内网相对快很多 - 打开vmdk文件，试了各种都不能打开，但是也许window系统下有工具能打开
解决方案二：创建虚拟机加载vmdk为第二硬盘
【创建虚拟机】，虚拟机需要自定义分区，主要修改卷组（VG）名称，修改卷组名称，主要是防止逻辑卷名称冲突
启动新的虚拟机，启动成功后，【添加硬盘】，硬盘文件选择需要打开的vmdk文件
进入虚拟机，执行下面命令
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27#检查激活状态
#有冲突就不会激活
[root@localhost ~]# lvscan
inactive '/dev/centos/swap' [15.75 GiB] inherit
inactive '/dev/centos/home' [1.93 TiB] inherit
inactive '/dev/centos/root' [50.00 GiB] inherit
ACTIVE '/dev/centos/swap' [2.00 GiB] inherit
ACTIVE '/dev/centos/home' [346.99 GiB] inherit
ACTIVE '/dev/centos/root' [50.00 GiB] inherit
#没有冲突会自动激活
[root@localhost ~]# lvscan
ACTIVE '/dev/centos/swap' [15.75 GiB] inherit
ACTIVE '/dev/centos/home' [1.93 TiB] inherit
ACTIVE '/dev/centos/root' [50.00 GiB] inherit
ACTIVE '/dev/os/swap' [2.00 GiB] inherit
ACTIVE '/dev/os/home' [346.99 GiB] inherit
ACTIVE '/dev/os/root' [50.00 GiB] inherit
#（可选不建议）如果忘了修改卷组（VG）名称，home目录可以在虚拟机里面修改，其他目录不行
umount /dev/centos/home #卸载
lvrename /dev/centos/home /dev/centos/home_old #重命名
mount /dev/centos/home_old /home
#挂载需要打开的vmdk之后执行下面的，激活命令
lvchange -ay /dev/centos/home
#--可选不建议结束---------
#挂载目录
mount /dev/centos/home /mnt/home
#然后就可以进行拷贝备份了
VMware ESXI常用基础操作
创建快照：选择虚拟机，点击
Actions>Snapshot>Take Snapshot。查看vmdk文件：在左侧菜单，点击
Storage>Data_disk>对应的虚拟机服务1
2
3
4#关键文件内容如下
server.vmdk # 基础文件
server-000001.vmdk #创建快照产生，从基础文件开始记录所有新的更改
server-000002.vmdk #纪录新的改变开启ESXI服务器的ssh功能：在左侧菜单，点击
Host>Manage>Services，里面开启ssh功能。开启后用户名和密码同网页登录esxi界面一样创建虚拟机：在左侧菜单，点击
Virtual Machines>create，选择linux，centos7，在iso里面选择Storage里面的iso镜像，然后就是安装centos的步骤了。设置静态ip和ESXI服务器在同一网段中。添加硬盘：选择虚拟机，点击
edit>add hard disk进行添加硬盘。合并快照磁盘：选择虚拟机，点击
Actions>Snapshot>Snapshot Manager>Consolidate。
Docker安装使用ES
安装
elasticsearch安装
在docker.elastic.co仓库查找springboot对应的es版本,，eg:
1
2
3
4# springboot 2.7.18
docker pull docker.elastic.co/elasticsearch/elasticsearch:7.17.22
#如果上面的拉取太慢了，可以用下面的dockerhub的镜像
docker pull elastic/elasticsearch:7.17.22创建配置文件，执行
vim /workspace/elasticsearch.yml写入如下内容1
2
3
4
5
6cluster.name: "docker-cluster"
network.host: 0.0.0.0
http.port: 9200
http.cors.enabled: true
http.cors.allow-origin: "*"
xpack.security.enabled: false-
1
2
3
4
5
6
7mkdir /workspace/v_es_data/
chmod -R 777 /workspace/v_es_data/
# -m限制使用内存1g
docker run -d -m 1g -p 9200:9200 -e "discovery.type=single-node" --name es \
-v /workspace/v_es_data:/usr/share/elasticsearch/data \
-v /workspace/elasticsearch.yml:/usr/share/elasticsearch/config/elasticsearch.yml \
docker.elastic.co/elasticsearch/elasticsearch:7.17.22 测试，执行下面命令
1
2
3➜ ~ curl -X GET "127.0.0.1:9200/_cat/nodes?v=true&pretty"
ip heap.percent ram.percent cpu load_1m load_5m load_15m node.role master name
172.18.0.2 35 84 5 0.06 0.09 0.13 cdfhilmrstw * eeeedfa1ed88
kibana安装（没采用）
可视化查看工具，类似dbeaver，非必需
执行下面命令进行安装
1
2
3docker run -d --name kibana -p 5601:5601 \
-e "ELASTICSEARCH_HOSTS=http://127.0.0.1:9200" \
docker.elastic.co/kibana/kibana:7.17.22访问
http://localhost:5601/然后进行各种操作，比较复杂
本地采用es-client插件
数据同步方案
应用层同步
在应用层面进行数据同步，即在应用代码中同时写入 MySQL 和 Elasticsearch。
优点：依赖少，自定义同步逻辑。
缺点：数据一致性和可靠性方面存在一些挑战，特别是在高并发的场景下。
MySQL binlog(数据库日志)
借助于数据库日志（如 MySQL binlog）来捕获数据变化，并将其同步到 Elasticsearch。这种方法通常能够实时同步数据，并且不会对数据库性能造成太大影响。
优点：实时同步数据，不会对数据库性能造成太大影响。
缺点：性能开销，binlog 会占用额外的磁盘空间。
Debezium(重量级别)
优点：功能强大，支持复杂的配置和扩展，社区活跃。
缺点：需要 Kafka，配置相对复杂。
Maxwell(轻量级别)
Maxwell是一种开源的MySQL数据库同步工具，它可以将MySQL数据库的binlog转化为JSON格式，并将其发送到消息队列中。
需要mq和kafka等中间件
优点：轻量级，配置简单，适合快速实现。
缺点：功能相对简单，不支持复杂的场景
canal
Canal是阿里巴巴开发的一款数据库同步工具，它可以实现MySQL数据库的binlog解析和日志抓取。模拟mysql的从节点实现。
springboot接入ES
在pom.xml添加依赖
1
2
3
4<dependency>
<groupId>org.springframework.boot</groupId>
<artifactId>spring-boot-starter-data-elasticsearch</artifactId>
</dependency>
maxwell
1 | version: '3.8' |
参考
Linux-install-cuda
centos7如何安装CUDA
查看显卡型号
lspci | grep -i vga，提示-bash: lspci: 未找到命令，执行yum install pciutils安装lspci命令1
2[root@exxk ~]# lspci | grep -i vga
01:00.0 VGA compatible controller: NVIDIA Corporation Device 25e2 (rev a1)在输出结果中获取
25e2关键字，在https://admin.pci-ids.ucw.cz/read/PC/10de/连接尾部拼接25e2，例如https://admin.pci-ids.ucw.cz/read/PC/10de/25e2，访问即可，页面回展示类似内容1
2
3Main -> PCI Devices -> Vendor 10de -> Device 10de:25e2
Name: GA107BM [GeForce RTX 3050 Mobile]从中可以知道25e2对应的显卡是
GeForce RTX 3050 Mobile在该https://www.nvidia.cn/Download/index.aspx?lang=cn网页找到`GeForce RTX 3050 Mobile
=GeForce RTX 3050 Laptop GPU该驱动，选择Linux64-bit然后下载下来NVIDIA-Linux-x86_64-550.78.run `拷贝到centos7上面，执行下面命令进行安装驱动
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67#屏蔽系统自带的nouveau显卡驱动（重启后生效---begin-----------
[root@exxk ~]# lsmod | grep nouveau
nouveau 2543616 0
drm_ttm_helper 12288 1 nouveau
ttm 90112 2 drm_ttm_helper,nouveau
i2c_algo_bit 12288 1 nouveau
mxm_wmi 12288 1 nouveau
drm_display_helper 176128 1 nouveau
drm_kms_helper 225280 4 drm_display_helper,nouveau
drm 643072 6 drm_kms_helper,drm_display_helper,drm_ttm_helper,ttm,nouveau
video 69632 2 ideapad_laptop,nouveau
wmi 36864 5 video,wmi_bmof,ideapad_laptop,mxm_wmi,nouveau
[root@exxk ~]# vi /lib/modprobe.d/dist-blacklist.conf
#将nvidiafb注释掉:
#blacklist nvidiafb
#然后添加以下语句：
#blacklist nouveau
#options nouveau modeset=0
[root@exxk ~]# mv /boot/initramfs-$(uname -r).img /boot/initramfs-$(uname -r).img.bak
[root@exxk ~]# dracut /boot/initramfs-$(uname -r).img $(uname -r)
[root@exxk ~]# reboot
[root@exxk ~]# lsmod | grep nouveau
#重启之后发现lsmod | grep nouveau命令无任何输出代表成功
#屏蔽系统自带的nouveau显卡驱动（重启后生效---end-----------
#安装依赖环境------------begin--------------------------
#安装kernel-devel kernel-headers gcc
[root@exxk ~]# yum install kernel-devel gcc -y
[root@exxk ~]# uname -r
3.10.0-1160.118.1.el7.x86_64
[root@exxk ~]# yum info kernel-devel kernel-headers | grep 发布
发布 ：1160.118.1.el7
发布 ：1160.118.1.el7
#内核版本要和安装的版本一致
#如果不一致，检查内核是否需要升级
[root@exxk ~]# yum check-update kernel
#查看可用内核
[root@exxk ~]# cat /boot/grub2/grub.cfg |grep menuentry
#升级到和安装工具一致的内核版本
[root@exxk ~]# grub2-set-default 'CentOS Linux (3.10.0-1160.118.1.el7.x86_64) 7 (Core)'
[root@exxk ~]# reboot
#安装依赖环境------------end--------------------------
#安装显卡驱动，一致点同意即可
[root@exxk ~]# ./NVIDIA-Linux-x86_64-550.78.run
#查看驱动信息
[root@exxk ~]# nvidia-smi
Sat May 11 14:43:05 2024
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 550.78 Driver Version: 550.78 CUDA Version: 12.4 |
|-----------------------------------------+------------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+========================+======================|
| 0 NVIDIA GeForce RTX 3050 ... Off | 00000000:01:00.0 Off | N/A |
| N/A 47C P0 10W / 60W | 1MiB / 4096MiB | 0% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+
+-----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| No running processes found |
+-----------------------------------------------------------------------------------------+安装NVIDIA Container Toolkit，执行下面的命令(确保您已安装适用于您的 Linux 发行版的NVIDIA 驱动程序 请注意，您不需要在主机系统上安装 CUDA Toolkit，但需要安装 NVIDIA 驱动程序)
1
2
3
4[root@exxk ~]# curl -s -L https://nvidia.github.io/libnvidia-container/stable/rpm/nvidia-container-toolkit.repo | sudo tee /etc/yum.repos.d/nvidia-container-toolkit.repo
[root@exxk ~]# yum-config-manager --enable nvidia-container-toolkit-experimental
yum install -y nvidia-container-toolkit
[root@exxk ~]# yum install -y nvidia-container-toolkit配置containerd（针对Kubernetes），执行下面命令
1
2
3
4
5[root@exxk ~]# nvidia-ctk runtime configure --runtime=containerd
INFO[0000] Loading config from /etc/containerd/config.toml
INFO[0000] Wrote updated config to /etc/containerd/config.toml
INFO[0000] It is recommended that containerd daemon be restarted.
[root@exxk ~]# systemctl restart containerd-
1
nvidia-docker pull registry.baidubce.com/paddlepaddle/paddle:2.6.1-gpu-cuda12.0-cudnn8.9-trt8.6
新版cuda安装(废弃采用容器安装)
安装cuda 12.4.1最新版(飞浆需要12.0的版本，因此将下面的替换成12.0.1)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33wget https://developer.download.nvidia.com/compute/cuda/12.4.1/local_installers/cuda_12.4.1_550.54.15_linux.run
sh cuda_12.4.1_550.54.15_linux.run
┌──────────────────────────────────────────────────────────────────────────────┐
│ CUDA Installer │
│ - [ ] Driver │
│ [ ] 550.54.15 │
│ + [X] CUDA Toolkit 12.4 │
│ [X] CUDA Demo Suite 12.4 │
│ [X] CUDA Documentation 12.4 │
│ - [ ] Kernel Objects │
│ [ ] nvidia-fs │
│ Options │
│ Install │
│ │
│ Up/Down: Move | Left/Right: Expand | 'Enter': Select | 'A': Advanced options │
└──────────────────────────────────────────────────────────────────────────────┘
===========
= Summary =
===========
Driver: Not Selected
Toolkit: Installed in /usr/local/cuda-12.4/
Please make sure that
- PATH includes /usr/local/cuda-12.4/bin
- LD_LIBRARY_PATH includes /usr/local/cuda-12.4/lib64, or, add /usr/local/cuda-12.4/lib64 to /etc/ld.so.conf and run ldconfig as root
To uninstall the CUDA Toolkit, run cuda-uninstaller in /usr/local/cuda-12.4/bin
***WARNING: Incomplete installation! This installation did not install the CUDA Driver. A driver of version at least 550.00 is required for CUDA 12.4 functionality to work.
To install the driver using this installer, run the following command, replacing <CudaInstaller> with the name of this run file:
sudo <CudaInstaller>.run --silent --driver
Logfile is /var/log/cuda-installer.log添加环境变量，修改
/etc/profile文件，执行vi /etc/profile，增加如下内容，然后执行source /etc/profile1
2export PATH=${PATH}:/usr/local/cuda-12.4/bin
export LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/usr/local/cuda-12.4/lib64验证
1
2
3
4
5
6[root@exxk ~]# nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2024 NVIDIA Corporation
Built on Thu_Mar_28_02:18:24_PDT_2024
Cuda compilation tools, release 12.4, V12.4.131
Build cuda_12.4.r12.4/compiler.34097967_0卸载
1
2
3
4
5
6
7
8
9
10
11
12
13
14[root@exxk ~]# cd /usr/local/cuda-12.4/bin
[root@exxk bin]# ./cuda-uninstaller
┌──────────────────────────────────────────────────────────────────────────────┐
│ CUDA Uninstaller │
│ [X] CUDA_Toolkit_12.4 │
│ [X] CUDA_Demo_Suite_12.4 │
│ [X] CUDA_Documentation_12.4 │
│ Done │
│ │
│ Up/Down: Move | 'Enter': Select │
└──────────────────────────────────────────────────────────────────────────────┘
Successfully uninstalled
[root@exxk local]# cd /usr/local/
[root@exxk local]# rm -rf cuda-12.4
旧版cuda废弃
下载 developer.nvidia.com页面按自己的环境选择，然后下载cuda。安装方式我选择的
runfile(local)，下载完成后上传到centos执行
sh cuda_7.5.18_linux.run，按ctrl+c跳过文档阅读1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47[root@exxk ~]# sh cuda_7.5.18_linux.run
Logging to /tmp/cuda_install_22223.log
Using more to view the EULA.
End User License Agreement
--------------------------
--More--(0%)
Do you accept the previously read EULA? (accept/decline/quit): accept
# 非常关键，我们已经在之前安装了高版本的驱动，这个千万别装。
Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 352.39? ((y)es/(n)o/(q)uit): n
Install the CUDA 7.5 Toolkit? ((y)es/(n)o/(q)uit): y
Enter Toolkit Location [ default is /usr/local/cuda-7.5 ]:
Do you want to install a symbolic link at /usr/local/cuda? ((y)es/(n)o/(q)uit): y
Install the CUDA 7.5 Samples? ((y)es/(n)o/(q)uit): y
Enter CUDA Samples Location [ default is /root ]:
Installing the CUDA Toolkit in /usr/local/cuda-7.5 ...
Missing recommended library: libGLU.so
Missing recommended library: libX11.so
Missing recommended library: libXi.so
Missing recommended library: libXmu.so
Installing the CUDA Samples in /root ...
Copying samples to /root/NVIDIA_CUDA-7.5_Samples now...
Finished copying samples.
===========
= Summary =
===========
Driver: Not Selected
Toolkit: Installed in /usr/local/cuda-7.5
Samples: Installed in /root, but missing recommended libraries
Please make sure that
- PATH includes /usr/local/cuda-7.5/bin
- LD_LIBRARY_PATH includes /usr/local/cuda-7.5/lib64, or, add /usr/local/cuda-7.5/lib64 to /etc/ld.so.conf and run ldconfig as root
To uninstall the CUDA Toolkit, run the uninstall script in /usr/local/cuda-7.5/bin
To uninstall the NVIDIA Driver, run nvidia-uninstall
Please see CUDA_Installation_Guide_Linux.pdf in /usr/local/cuda-7.5/doc/pdf for detailed information on setting up CUDA.
***WARNING: Incomplete installation! This installation did not install the CUDA Driver. A driver of version at least 352.00 is required for CUDA 7.5 functionality to work.
To install the driver using this installer, run the following command, replacing <CudaInstaller> with the name of this run file:
sudo <CudaInstaller>.run -silent -driver
Logfile is /tmp/cuda_install_22223.log
Signal caught, cleaning up添加环境变量，修改
/etc/profile文件，执行vi /etc/profile，增加如下内容，然后执行source /etc/profile1
2export PATH=${PATH}:/usr/local/cuda/bin
export LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/usr/local/cuda/lib64验证环境
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57[root@exxk ~]# nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2015 NVIDIA Corporation
Built on Tue_Aug_11_14:27:32_CDT_2015
Cuda compilation tools, release 7.5, V7.5.17
# 如果前面安装了CUDA的example，这里可以执行如下操作：
[root@exxk ~]# cd /root/NVIDIA_CUDA-7.5_Samples/1_Utilities/deviceQuery
[root@exxk deviceQuery]# yum install gcc gcc-c++
[root@exxk deviceQuery]# make
"/usr/local/cuda-7.5"/bin/nvcc -ccbin g++ -I../../common/inc -m64 -gencode arch=compute_20,code=sm_20 -gencode arch=compute_30,code=sm_30 -gencode arch=compute_35,code=sm_35 -gencode arch=compute_37,code=sm_37 -gencode arch=compute_50,code=sm_50 -gencode arch=compute_52,code=sm_52 -gencode arch=compute_52,code=compute_52 -o deviceQuery.o -c deviceQuery.cpp
"/usr/local/cuda-7.5"/bin/nvcc -ccbin g++ -m64 -gencode arch=compute_20,code=sm_20 -gencode arch=compute_30,code=sm_30 -gencode arch=compute_35,code=sm_35 -gencode arch=compute_37,code=sm_37 -gencode arch=compute_50,code=sm_50 -gencode arch=compute_52,code=sm_52 -gencode arch=compute_52,code=compute_52 -o deviceQuery deviceQuery.o
mkdir -p ../../bin/x86_64/linux/release
cp deviceQuery ../../bin/x86_64/linux/release
[root@exxk deviceQuery]# ./deviceQuery
./deviceQuery Starting...
CUDA Device Query (Runtime API) version (CUDART static linking)
Detected 1 CUDA Capable device(s)
Device 0: "NVIDIA GeForce RTX 3050 Laptop GPU"
CUDA Driver Version / Runtime Version 12.4 / 7.5
CUDA Capability Major/Minor version number: 8.6
Total amount of global memory: 3873 MBytes (4060872704 bytes)
MapSMtoCores for SM 8.6 is undefined. Default to use 128 Cores/SM
MapSMtoCores for SM 8.6 is undefined. Default to use 128 Cores/SM
(16) Multiprocessors, (128) CUDA Cores/MP: 2048 CUDA Cores
GPU Max Clock rate: 1500 MHz (1.50 GHz)
Memory Clock rate: 6001 Mhz
Memory Bus Width: 128-bit
L2 Cache Size: 1572864 bytes
Maximum Texture Dimension Size (x,y,z) 1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)
Maximum Layered 1D Texture Size, (num) layers 1D=(32768), 2048 layers
Maximum Layered 2D Texture Size, (num) layers 2D=(32768, 32768), 2048 layers
Total amount of constant memory: 65536 bytes
Total amount of shared memory per block: 49152 bytes
Total number of registers available per block: 65536
Warp size: 32
Maximum number of threads per multiprocessor: 1536
Maximum number of threads per block: 1024
Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
Max dimension size of a grid size (x,y,z): (2147483647, 65535, 65535)
Maximum memory pitch: 2147483647 bytes
Texture alignment: 512 bytes
Concurrent copy and kernel execution: Yes with 2 copy engine(s)
Run time limit on kernels: No
Integrated GPU sharing Host Memory: No
Support host page-locked memory mapping: Yes
Alignment requirement for Surfaces: Yes
Device has ECC support: Disabled
Device supports Unified Addressing (UVA): Yes
Device PCI Domain ID / Bus ID / location ID: 0 / 1 / 0
Compute Mode:
< Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >
deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 12.4, CUDA Runtime Version = 7.5, NumDevs = 1, Device0 = NVIDIA GeForce RTX 3050 Laptop GPU
Result = PASS卸载
/usr/local/cuda-7.5/bin/uninstall_cuda_7.5.pl
参考
AI-PaddleDetection
PaddleDetection是一个基于PaddlePaddle的目标检测端到端开发套件。
PP-Human 实时行人分析工具
安装使用步骤
在PyCharm欢迎页通过Get from VCS拉取github上PaddlePaddle/PaddleSeg项目。
拉取成功后，PyCharm会提示创建venv虚拟环境，点击确定即可，在PyCharm打开终端命令行前面有venv代表创建成功，如果没有，关闭终端点击
Add new Interpreter（需要python3.10的环境），添加后，重新在PyCharm打开终端1
2例如:
(venv) ➜ PaddleSeg git:(release/2.9)安装paddle，在PyCharm中Terminal命令行中执行
pip install paddlepaddle安装setuptools，在PyCharm中Terminal命令行中执行
pip install -U pip setuptools安装依赖，在PyCharm中Terminal命令行中执行
pip install -r requirements.txt
使用步骤
1 | # 运行图片检测人，结果输出在output目录 |
检查人并输出视频
安装ffmpeg：
brew install ffmpeg下载安装rtsp推流服务bluenviron/mediamtx，在release页面选择适合的版本，我的是inter的mac，选择mediamtx_v1.8.1_darwin_amd64.tar.gz即可
解压，然后运行
./mediamtx运行视频流检测并推流
python deploy/pipeline/pipeline.py --config deploy/pipeline/config/examples/infer_cfg_human_attr.yml --rtsp rtsp://admin:123456@172.1.1.1:554/ch01.264?dev=1 --pushurl rtsp://127.0.0.1:85541
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69python deploy/pipeline/pipeline.py --config deploy/pipeline/config/examples/infer_cfg_human_attr.yml --rtsp rtsp://admin:123456@172.16.80.59:554/ch01.264?dev=1 --pushurl rtsp://127.0.0.1:8554
----------- Running Arguments -----------
ATTR:
batch_size: 8
enable: true
model_dir: https://bj.bcebos.com/v1/paddledet/models/pipeline/PPLCNet_x1_0_person_attribute_945_infer.zip
DET:
batch_size: 1
model_dir: https://bj.bcebos.com/v1/paddledet/models/pipeline/pphuman/ppyoloe_plus_crn_t_auxhead_320_60e_pphuman.tar.gz
MOT:
batch_size: 1
enable: true
model_dir: https://bj.bcebos.com/v1/paddledet/models/pipeline/pphuman/ppyoloe_plus_crn_t_auxhead_320_60e_pphuman.tar.gz
tracker_config: deploy/pipeline/config/tracker_config.yml
attr_thresh: 0.5
crop_thresh: 0.5
visual: true
warmup_frame: 50
------------------------------------------
Multi-Object Tracking enabled
Human Attribute Recognition enabled
100%|██████████| 15897/15897 [00:05<00:00, 2768.11KB/s]
DET model dir: /.cache/paddle/infer_weights/ppyoloe_plus_crn_t_auxhead_320_60e_pphuman
MOT model dir: /.cache/paddle/infer_weights/ppyoloe_plus_crn_t_auxhead_320_60e_pphuman
ATTR model dir: /.cache/paddle/infer_weights/PPLCNet_x1_0_person_attribute_945_infer
----------- Model Configuration -----------
Model Arch: StrongBaseline
Transform Order:
--transform op: Resize
--transform op: NormalizeImage
--transform op: Permute
--------------------------------------------
----------- Model Configuration -----------
Model Arch: PPYOLOE
Transform Order:
--transform op: Resize
--transform op: Permute
--------------------------------------------
video fps: 25, frame_count: -2562047788015215
the result will push stream to url:rtsp://127.0.0.1:8554/ch01
ffmpeg version 7.0 Copyright (c) 2000-2024 the FFmpeg developers
built with Apple clang version 15.0.0 (clang-1500.1.0.2.5)
configuration: --prefix=/usr/local/Cellar/ffmpeg/7.0_1 --enable-shared --enable-pthreads --enable-version3 --cc=clang --host-cflags= --host-ldflags='-Wl,-ld_classic' --enable-ffplay --enable-gnutls --enable-gpl --enable-libaom --enable-libaribb24 --enable-libbluray --enable-libdav1d --enable-libharfbuzz --enable-libjxl --enable-libmp3lame --enable-libopus --enable-librav1e --enable-librist --enable-librubberband --enable-libsnappy --enable-libsrt --enable-libssh --enable-libsvtav1 --enable-libtesseract --enable-libtheora --enable-libvidstab --enable-libvmaf --enable-libvorbis --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxml2 --enable-libxvid --enable-lzma --enable-libfontconfig --enable-libfreetype --enable-frei0r --enable-libass --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-libopenvino --enable-libspeex --enable-libsoxr --enable-libzmq --enable-libzimg --disable-libjack --disable-indev=jack --enable-videotoolbox --enable-audiotoolbox
libavutil 59. 8.100 / 59. 8.100
libavcodec 61. 3.100 / 61. 3.100
libavformat 61. 1.100 / 61. 1.100
libavdevice 61. 1.100 / 61. 1.100
libavfilter 10. 1.100 / 10. 1.100
libswscale 8. 1.100 / 8. 1.100
libswresample 5. 1.100 / 5. 1.100
libpostproc 58. 1.100 / 58. 1.100
Thread: 0; frame id: 0
Thread: 0; trackid number: 2
Input #0, rawvideo, from 'fd:':
Duration: N/A, start: 0.000000, bitrate: 552960 kb/s
Stream #0:0: Video: rawvideo (BGR[24] / 0x18524742), bgr24, 1280x720, 552960 kb/s, 25 tbr, 25 tbn
Stream mapping:
Stream #0:0 -> #0:0 (rawvideo (native) -> mpeg4 (native))
Output #0, rtsp, to 'rtsp://127.0.0.1:8554/ch01':
Metadata:
encoder : Lavf61.1.100
Stream #0:0: Video: mpeg4, yuv420p(progressive), 1280x720, q=2-31, 200 kb/s, 25 fps, 90k tbn
Metadata:
encoder : Lavc61.3.100 mpeg4
Side data:
cpb: bitrate max/min/avg: 0/0/200000 buffer size: 0 vbv_delay: N/A
Thread: 0; frame id: 10
Thread: 0; trackid number: 2然后使用vlc播放
rtsp://127.0.0.1:8554/ch01流即可，流的地址在上一步日志里面可以看到the result will push stream to url:rtsp://127.0.0.1:8554/ch01默认的模型，识别速度个人电脑会卡顿，可以替换为小模型，模型地址可以在PP-Human快速开始-模型下载里面进行找
1
2
3
4
5修改该deploy/pipeline/config/examples/infer_cfg_human_attr.yml文件
替换
model_dir: https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_l_36e_pipeline.zip
为
model_dir: https://bj.bcebos.com/v1/paddledet/models/pipeline/pphuman/ppyoloe_plus_crn_t_auxhead_320_60e_pphuman.tar.gz
常见错误
安装依赖，执行
pip install -r requirements.txt时报AttributeError: module 'pkgutil' has no attribute 'ImpImporter'. Did you mean: 'zipimporter'?错误，详细信息如下：1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54(.venv) ➜ PaddleDetection git:(release/2.7) pip install -r requirements.txt
Collecting numpy<1.24 (from -r requirements.txt (line 1))
Using cached numpy-1.23.5.tar.gz (10.7 MB)
Installing build dependencies ... done
Getting requirements to build wheel ... error
error: subprocess-exited-with-error
× Getting requirements to build wheel did not run successfully.
│ exit code: 1
╰─> [33 lines of output]
Traceback (most recent call last):
File "workspace/PaddleDetection/.venv/lib/python3.12/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 353, in <module>
main()
File "workspace/PaddleDetection/.venv/lib/python3.12/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 335, in main
json_out['return_val'] = hook(**hook_input['kwargs'])
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "workspace/PaddleDetection/.venv/lib/python3.12/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 112, in get_requires_for_build_wheel
backend = _build_backend()
^^^^^^^^^^^^^^^^
File "workspace/PaddleDetection/.venv/lib/python3.12/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 77, in _build_backend
obj = import_module(mod_path)
^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/Cellar/python@3.12/3.12.2_1/Frameworks/Python.framework/Versions/3.12/lib/python3.12/importlib/__init__.py", line 90, in import_module
return _bootstrap._gcd_import(name[level:], package, level)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "<frozen importlib._bootstrap>", line 1387, in _gcd_import
File "<frozen importlib._bootstrap>", line 1360, in _find_and_load
File "<frozen importlib._bootstrap>", line 1310, in _find_and_load_unlocked
File "<frozen importlib._bootstrap>", line 488, in _call_with_frames_removed
File "<frozen importlib._bootstrap>", line 1387, in _gcd_import
File "<frozen importlib._bootstrap>", line 1360, in _find_and_load
File "<frozen importlib._bootstrap>", line 1331, in _find_and_load_unlocked
File "<frozen importlib._bootstrap>", line 935, in _load_unlocked
File "<frozen importlib._bootstrap_external>", line 995, in exec_module
File "<frozen importlib._bootstrap>", line 488, in _call_with_frames_removed
File "/private/var/folders/jd/0572r10n2rbcwn4tf1w6sy500000gn/T/pip-build-env-r6d8n8j4/overlay/lib/python3.12/site-packages/setuptools/__init__.py", line 16, in <module>
import setuptools.version
File "/private/var/folders/jd/0572r10n2rbcwn4tf1w6sy500000gn/T/pip-build-env-r6d8n8j4/overlay/lib/python3.12/site-packages/setuptools/version.py", line 1, in <module>
import pkg_resources
File "/private/var/folders/jd/0572r10n2rbcwn4tf1w6sy500000gn/T/pip-build-env-r6d8n8j4/overlay/lib/python3.12/site-packages/pkg_resources/__init__.py", line 2172, in <module>
register_finder(pkgutil.ImpImporter, find_on_path)
^^^^^^^^^^^^^^^^^^^
AttributeError: module 'pkgutil' has no attribute 'ImpImporter'. Did you mean: 'zipimporter'?
[end of output]
note: This error originates from a subprocess, and is likely not a problem with pip.
error: subprocess-exited-with-error
× Getting requirements to build wheel did not run successfully.
│ exit code: 1
╰─> See above for output.
note: This error originates from a subprocess, and is likely not a problem with pip.原因：在
docs/tutorials/INSTALL_cn.md文档里面可以查看到只支持Python 3(3.5.1+/3.6/3.7/3.8/3.9/3.10)，64位版本。解决：执行
brew install python@3.10，然后在上面安装的第二个步骤里面，切换成3.10的python环境出现提示安装
pip install numba==0.56.4的警告，安装即可运行视频流检测人时，报错
1
2
3
4
5
6
7
8
9
10
11
12
13python deploy/pipeline/pipeline.py --config deploy/pipeline/config/examples/infer_cfg_human_attr.yml -o visual=False --rtsp rtsp://admin:123456@172.16.1.1:554/ch01.264?dev=1
Traceback (most recent call last):
File "workspace/PaddleDetection/deploy/pipeline/pipeline.py", line 1321, in <module>
main()
File "workspace/PaddleDetection/deploy/pipeline/pipeline.py", line 1303, in main
cfg = merge_cfg(FLAGS) # use command params to update config
File "workspace/PaddleDetection/deploy/pipeline/cfg_utils.py", line 212, in merge_cfg
pred_config = merge_opt(pred_config, args_dict)
File "workspace/PaddleDetection/deploy/pipeline/cfg_utils.py", line 202, in merge_opt
for sub_k, sub_v in value.items():
AttributeError: 'bool' object has no attribute 'items'
Process finished with exit code 1解决：删除命令里面的
-o visual=False，似乎是该参数有问题运行视频流检测加推流的时候，报错
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60python deploy/pipeline/pipeline.py --config deploy/pipeline/config/examples/infer_cfg_human_attr.yml --rtsp rtsp://admin:iexxk123456@172.16.80.229:554/ch01.264?dev=1 --pushurl rtsp://127.0.0.1:8554
----------- Running Arguments -----------
ATTR:
batch_size: 8
enable: true
model_dir: https://bj.bcebos.com/v1/paddledet/models/pipeline/PPLCNet_x1_0_person_attribute_945_infer.zip
DET:
batch_size: 1
model_dir: https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_l_36e_pipeline.zip
MOT:
batch_size: 1
enable: true
model_dir: https://bj.bcebos.com/v1/paddledet/models/pipeline/mot_ppyoloe_l_36e_pipeline.zip
tracker_config: deploy/pipeline/config/tracker_config.yml
attr_thresh: 0.5
crop_thresh: 0.5
visual: true
warmup_frame: 50
------------------------------------------
Multi-Object Tracking enabled
Human Attribute Recognition enabled
DET model dir: .cache/paddle/infer_weights/mot_ppyoloe_l_36e_pipeline
MOT model dir: .cache/paddle/infer_weights/mot_ppyoloe_l_36e_pipeline
ATTR model dir: .cache/paddle/infer_weights/PPLCNet_x1_0_person_attribute_945_infer
----------- Model Configuration -----------
Model Arch: StrongBaseline
Transform Order:
--transform op: Resize
--transform op: NormalizeImage
--transform op: Permute
--------------------------------------------
----------- Model Configuration -----------
Model Arch: YOLO
Transform Order:
--transform op: Resize
--transform op: Permute
--------------------------------------------
video fps: 25, frame_count: -2562047788015215
the result will push stream to url:rtsp://127.0.0.1:8554/ch01
Traceback (most recent call last):
File "workspace/PaddleDetection/deploy/pipeline/pipeline.py", line 1321, in <module>
main()
File "workspace/PaddleDetection/deploy/pipeline/pipeline.py", line 1308, in main
pipeline.run_multithreads()
File "workspace/PaddleDetection/deploy/pipeline/pipeline.py", line 179, in run_multithreads
self.predictor.run(self.input)
File "workspace/PaddleDetection/deploy/pipeline/pipeline.py", line 533, in run
self.predict_video(input, thread_idx=thread_idx)
File "workspace/PaddleDetection/deploy/pipeline/pipeline.py", line 670, in predict_video
pushstream.initcmd(fps, width, height)
File "workspace/PaddleDetection/deploy/pipeline/pipe_utils.py", line 142, in initcmd
self.pipe = sp.Popen(self.command, stdin=sp.PIPE)
File "/usr/local/Cellar/python@3.10/3.10.14/Frameworks/Python.framework/Versions/3.10/lib/python3.10/subprocess.py", line 971, in __init__
self._execute_child(args, executable, preexec_fn, close_fds,
File "/usr/local/Cellar/python@3.10/3.10.14/Frameworks/Python.framework/Versions/3.10/lib/python3.10/subprocess.py", line 1863, in _execute_child
raise child_exception_type(errno_num, err_msg, err_filename)
FileNotFoundError: [Errno 2] No such file or directory: 'ffmpeg'
Process finished with exit code 1解决：安装ffmpeg，执行
brew install ffmpeg
Tools-NVR
Agent DVR
下载安装即可
zoneminder
wsl2环境安装（Ubuntu22版本安装手册）
执行下面的安装命令
1
2
3
4
5
6
7
8
9
10sudo apt-get update && sudo apt upgrade -y
sudo apt install -y software-properties-common
sudo add-apt-repository ppa:iconnor/zoneminder-1.36
sudo apt update
sudo apt install -y zoneminder
sudo a2enmod rewrite
sudo a2enconf zoneminder
sudo systemctl restart apache2
sudo systemctl enable zoneminder
sudo systemctl start zoneminder访问
http://hostname_or_ip/zm添加监视器，查看官方使用手册
测试
两个都无法正常用下面脚本获取摄像头信息
测试脚本nvr_test.py
1 | # pip install onvif_zeep |
AI基础知识
常见的AI深度学习框架
PaddleSeg
飞桨（PaddlePaddle）以百度多年的深度学习技术研究和业务应用为基础，集深度学习核心训练和推理框架、基础模型库、端到端开发套件、丰富的工具组件于一体，是中国首个自主研发、功能完备、开源开放的产业级深度学习平台。国内的可以优先选择paddlepaddle，这是因为百度在大力推广该框架，能提供丰富的算力支持和技术支持，且百度开源了众多模型和应用，是人工智能入门的一个好选择，且能在模型上面进行一系列魔改，强烈推荐！
pytorch
是由Facebook人工智能学院提供支持服务的，目前主要在学术研究方向领域处于领先地位，许多学术论文都是用pytorch编写的，因此使用范围更广
常见的标注工具
labelImg
目前还没用过，官方已经不维护，建议用label-studio
label-studio
还没用过，docker安装
labelme
启动有点慢
常用数据集检索网站
优先尝试：
● 百度飞桨开源数据集
● OpenDataLab 引领AI大模型时代的开放数据平台
● kaggle开源数据集
● 谷歌开源数据集
● ModelScope开源数据集
● TIANCHI天池开源数据集
● HuggingFace开源数据集
● 微软开源数据集
● Roboflow Universe: Open Source Computer Vision Community
备选：
● 启智开源数据集
● 和鲸开源数据集
● openimages开源数据集
● GitHub开源数据集
● AWS亚马逊开源数据集
● LUGE千言开源数据集
● UCI开源数据集
● 计算机视觉开源数据集
● Dataju聚数力开源数据集
● Hyper超神经开源数据集
● BAAI开源数据集
注意事项：
1. 检索关键词
2. 数据集格式
PaddleSeg基于机器学习的图像分割
PyCharm安装PaddleSeg
在PyCharm欢迎页通过Get from VCS拉取github上PaddlePaddle/PaddleSeg项目
拉取成功后，PyCharm会提示创建venv虚拟环境，点击确定即可，在PyCharm打开终端命令行前面有venv代表创建成功，如果没有，关闭终端点击
Add new Interpreter，添加后，重新在PyCharm打开终端1
2例如:
(venv) ➜ PaddleSeg git:(release/2.9)安装paddle，在PyCharm中Terminal命令行中执行
pip install paddlepaddle安装setuptools，在PyCharm中Terminal命令行中执行
pip install -U pip setuptools[可选]，检查paddle是否安装成功，在PyCharm中Python console执行
import paddle和paddle.utils.run_check()，检查版本执行print(paddle.__version__)安装paddleseg，在PyCharm中Terminal命令行中执行
pip install paddleseg，(这里采用的直接安装发布的版本，本地编译未实验通过)验证安装是否成功，在在PyCharm中Terminal命令行中执行
sh tests/install/check_predict.sh
PaddleSeg使用
准备自定义数据集
标注工具(标注数据)
PddleSeg已支持2种标注工具：LabelMe、精灵数据标注工具
下载安装labelme
根据文档操作，进行标注数据，简单流程
打开图片目录->创建多边形->框选要标注的数据->保存会在图片目录生成一个json文件，标注目录下的所有图片1
2
3
4
5
6
7
8
9
10
11
12
13#标注前目录结构
paddle
|--image1.jpg
|--image2.jpg
|--...
#标注后目录结构
paddle
|--image1.jpg
|--image2.jpg
|--...
|--image1.json
|--image2.json
|--...将标注的数据转换为模型训练时所需的数据格式
1
2
3
4
5
6
7
8
9
10
11
12
13
14#python tools/data/labelme2seg.py [-h] 图片目录 输出目录
python tools/data/labelme2seg.py /Users/x/Downloads/paddle /Users/xuanleung/Downloads/paddle_pr
#转换后目录结构
paddle
paddle_pr
|--annotations #红色背景，绿色标注的图片
| |--image1.png
| |--image2.png
| |--...
|--images
| |--image1.jpg
| |--image2.jpg
| |--...
|--class_names.txt #是数据集中所有标注类别的名称
切分数据
切分数据，对于所有原始图像和标注图像，需要按照比例划分为训练集、验证集、测试集。
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20#python tools/data/split_dataset_list.py <dataset_root:原始图像目录名> <images_dir_name:原始图像目录名> <labels_dir_name:标注图像目录名> ${FLAGS}
python tools/data/split_dataset_list.py /Users/x/Downloads/paddle_pr images annotations --split 0.6 0.2 0.2 --format jpg png
#执行后目录结构
paddle_pr
|--test.txt #测试集
|--val.txt #验证集
|--train.txt #训练集
|--annotations #红色背景，绿色标注的图片
| |--image1.png
| |--image2.png
| |--...
|--images
| |--image1.jpg
| |--image2.jpg
| |--...
|--class_names.txt #是数据集中所有标注类别的名称
#txt文件内容格式
images/image1.jpg annotations/image1.png
images/image2.jpg annotations/image2.png
....FLAGS说明：
FLAG 含义 默认值 参数数目 –split 训练集、验证集和测试集的切分比例 0.7 0.3 0 3 –separator txt文件列表分隔符 “ “ 1 –format 原始图像和标注图像的图片后缀 “jpg” “png” 2 –postfix 按文件主名（无扩展名）是否包含指定后缀对图片和标签集进行筛选 “” “”（2个空字符） 2
准备配置文件
拷贝paddle_pr数据集目录到项目根目录，在项目下面的目录configs/quick_start/pp_liteseg_optic_disc_512x512_1k.yml，修改这两个路径配置：
1 | train_dataset: |
配置文件说明：
1 | batch_size: 4 #设定batch_size的值即为迭代一次送入网络的图片数量，一般显卡显存越大，batch_size的值可以越大。如果使用多卡训练，总得batch size等于该batch size乘以卡数。 |
训练模型
使用pyCharm添加一个python run config的启动配置，script设置为tools/train.py，script parameters设置为--config configs/quick_start/pp_liteseg_optic_disc_512x512_1k.yml --save_interval 500 --do_eval --use_vdl --save_dir output，然后点击右上角的run。
运行相关的日志如下：
1 | /Users/x/workspace/PaddleSeg/venv/bin/python -X pycache_prefix=/Users/x/Library/Caches/JetBrains/PyCharm2023.3/cpython-cache /Applications/PyCharm.app/Contents/plugins/python/helpers/pydev/pydevd.py --multiprocess --qt-support=auto --client 127.0.0.1 --port 64923 --file /Users/x/workspace/PaddleSeg/tools/train.py --config configs/quick_start/pp_liteseg_optic_disc_512x512_1k.yml --save_interval 500 --do_eval --use_vdl --save_dir output |
生成的文件：
1 | output |
模型评估
拷贝paddle_pr数据集目录到项目根目录，在项目下面的目录configs/quick_start/pp_liteseg_optic_disc_512x512_1k.yml，修改这两个路径配置：
1 | val_dataset: |
使用pyCharm添加一个python run config的启动配置，script设置为tools/val.py，script parameters设置为--config configs/quick_start/pp_liteseg_optic_disc_512x512_1k.yml --model_path output/iter_1000/model.pdparams，然后点击右上角的run。
输出结果：
1 | /Users/x/workspace/PaddleSeg/venv/bin/python tools/val.py --config configs/quick_start/pp_liteseg_optic_disc_512x512_1k.yml --model_path output/iter_1000/model.pdparams |
模型预测
使用pyCharm添加一个python run config的启动配置，script设置为tools/predict.py，script parameters设置为--config configs/quick_start/pp_liteseg_optic_disc_512x512_1k.yml --model_path output/iter_1000/model.pdparams --image_path paddle_pr/images/image1.jpg --save_dir output/result，然后点击右上角的run。
输出结果：
1 | /Users/x/workspace/PaddleSeg/venv/bin/python tools/predict.py --config configs/quick_start/pp_liteseg_optic_disc_512x512_1k.yml --model_path output/iter_1000/model.pdparams --image_path paddle_pr/images/image1.jpg --save_dir output/result |
生成的文件：
1 | output/result |
导出模型
使用pyCharm添加一个python run config的启动配置，script设置为tools/export.py，script parameters设置为--config configs/quick_start/pp_liteseg_optic_disc_512x512_1k.yml --model_path output/best_model/model.pdparams --save_dir output/inference_model，然后点击右上角的run。
输出结果：
1 | /Users/x/workspace/PaddleSeg/venv/bin/python tools/export.py --config configs/quick_start/pp_liteseg_optic_disc_512x512_1k.yml --model_path output/best_model/model.pdparams --save_dir output/inference_model |
生成的文件：
1 | output/inference_model |
部署模型(Paddle Inference部署python)
使用pyCharm添加一个python run config的启动配置，script设置为deploy/python/infer.py，script parameters设置为-config output/inference_model/deploy.yaml --image_path /Users/xuanleung/Downloads/test.jpg --device cpu，然后点击右上角的run。
生产的文件：output/test.png
MacOS安装
1 | #检查环境 |
macOS docker 安装
1 | mkdir paddle |
常见问题
安装
python3 -m pip install paddlepaddle==2.6.0 -i https://mirror.baidu.com/pypi/simple提示如下错误1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18➜ ~ python3 -m pip install paddlepaddle==2.6.0 -i https://mirror.baidu.com/pypi/simple
error: externally-managed-environment
× This environment is externally managed
╰─> To install Python packages system-wide, try brew install
xyz, where xyz is the package you are trying to
install.
If you wish to install a non-brew-packaged Python package,
create a virtual environment using python3 -m venv path/to/venv.
Then use path/to/venv/bin/python and path/to/venv/bin/pip.
If you wish to install a non-brew packaged Python application,
it may be easiest to use pipx install xyz, which will manage a
virtual environment for you. Make sure you have pipx installed.
note: If you believe this is a mistake, please contact your Python installation or OS distribution provider. You can override this, at the risk of breaking your Python installation or OS, by passing --break-system-packages.
hint: See PEP 668 for the detailed specification.解决：参考pip(3) install，完美解决 externally-managed-environment
方案一：添加参数–break-system-packages，这种直接安装到系统，可能会影响系统环境。
1
python3 -m pip install paddlepaddle==2.6.0 -i https://mirror.baidu.com/pypi/simple --break-system-packages
方案二：pipx，安装完成后，无法导入paddle，初步判断无法进入虚拟环境
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28# pipx会为每个安装的应用创建一个独立的虚拟环境，避免不同应用之间的依赖冲突。
➜ ~ brew install pipx
#当你首次安装pipx时，运行pipx ensurepath会自动检查并修改你的环境变量设置（如需要的话），以确保你可以轻松运行pipx安装的任何程序。这个步骤通常只需要执行一次。
➜ ~ pipx ensurepath
#替换官方的安装命令，使用pipx安装
# 2.9.0找不到
➜ ~ pipx install paddlepaddle==2.9.0 -i https://mirror.baidu.com/pypi/simple
Fatal error from pip prevented installation. Full pip output in file:
/Users/xuanleung/Library/Logs/pipx/cmd_2024-03-25_15.41.49_pip_errors.log
Some possibly relevant errors from pip install:
ERROR: Could not find a version that satisfies the requirement paddlepaddle==2.9.0 (from versions: 2.6.0)
ERROR: No matching distribution found for paddlepaddle==2.9.0
Error installing paddlepaddle from spec 'paddlepaddle==2.9.0'.
# 换成安装2.6.0
➜ ~ pipx install paddlepaddle==2.6.0 -i https://mirror.baidu.com/pypi/simple
installed package paddlepaddle 2.6.0, installed using Python 3.12.2
These apps are now globally available
- fleetrun
- paddle
⚠️ Note: '/Users/xuanleung/.local/bin' is not on your PATH environment
variable. These apps will not be globally accessible until your PATH is
updated. Run `pipx ensurepath` to automatically add it, or manually modify
your PATH in your shell's config file (i.e. ~/.bashrc).
done! ✨ 🌟 ✨
# 再次执行更新环境变量
➜ ~ pipx ensurepath【采用】方案三：使用venv
执行
import paddle提示Python 3: ImportError “No Module named Setuptools”1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21>>> import paddle
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/Users/xx/workspace/paddleseg_python3/lib/python3.12/site-packages/paddle/__init__.py", line 28, in <module>
from .base import core # noqa: F401
^^^^^^^^^^^^^^^^^^^^^^
File "/Users/xx/workspace/paddleseg_python3/lib/python3.12/site-packages/paddle/base/__init__.py", line 77, in <module>
from . import dataset
File "/Users/xx/workspace/paddleseg_python3/lib/python3.12/site-packages/paddle/base/dataset.py", line 20, in <module>
from ..utils import deprecated
File "/Users/xx/workspace/paddleseg_python3/lib/python3.12/site-packages/paddle/utils/__init__.py", line 16, in <module>
from . import ( # noqa: F401
File "/Users/xx/workspace/paddleseg_python3/lib/python3.12/site-packages/paddle/utils/cpp_extension/__init__.py", line 15, in <module>
from .cpp_extension import (
File "/Users/xx/workspace/paddleseg_python3/lib/python3.12/site-packages/paddle/utils/cpp_extension/cpp_extension.py", line 21, in <module>
import setuptools
ModuleNotFoundError: No module named 'setuptools'
>>> paddle.utils.run_check()
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
NameError: name 'paddle' is not defined解决：执行
pip install -U pip setuptools执行
1
2
3
4
5
6python tools/train.py \
--config configs/quick_start/pp_liteseg_optic_disc_512x512_1k.yml \
--do_eval \
--use_vdl \
--save_interval 500 \
--save_dir output提示如下错误:
1
2
3
4
5
6
7
8
9
10
11
12
13(venv) ➜ PaddleSeg git:(release/2.9) ✗ python tools/train.py \
--config configs/quick_start/pp_liteseg_optic_disc_512x512_1k.yml \
--save_interval 500 \
--do_eval \
--use_vdl \
--save_dir output
Traceback (most recent call last):
File "/Users/x/workspace/PaddleSeg/tools/train.py", line 213, in <module>
main(args)
File "/Users/x/workspace/PaddleSeg/tools/train.py", line 145, in main
cfg = Config(
^^^^^^^
TypeError: Config.__init__() got an unexpected keyword argument 'to_static_training'原因：
tools/train.py文件里面缺少参数to_static_training
解决：切换到release/2.8.1分支，再次执行命令执行
1
2
3
4
5
6python tools/train.py \
--config configs/quick_start/pp_liteseg_optic_disc_512x512_1k.yml \
--do_eval \
--use_vdl \
--save_interval 500 \
--save_dir output提示如下错误：
1
2
3
4
5
6
7
8
9
10
11
12
13
14- type: CrossEntropyLoss
- type: CrossEntropyLoss
- type: CrossEntropyLoss
/Users/xuanleung/workspace/PaddleSeg/venv/lib/python3.12/site-packages/paddle/nn/layer/norm.py:824: UserWarning: When training, we now always track global mean and variance.
warnings.warn(
Traceback (most recent call last):
File "/Users/x/workspace/PaddleSeg/tools/train.py", line 195, in <module>
main(args)
File "/Users/x/workspace/PaddleSeg/tools/train.py", line 170, in main
train(
File "/Users/x/workspace/PaddleSeg/venv/lib/python3.12/site-packages/paddleseg/core/train.py", line 273, in train
avg_loss_list = [l[0] / log_iters for l in avg_loss_list]
~^^^
IndexError: too many indices for array: array is 0-dimensional, but 1 were indexed解决：使用pyCharm添加一个
python run config的启动配置，script设置为tools/train.py，script parameters设置为--config configs/quick_start/pp_liteseg_optic_disc_512x512_1k.yml --save_interval 500 --do_eval --use_vdl --save_dir output，然后点击右上角的run。
批量从照片中提取文本
tesseract
识别率较低(已卸载)
安装
mac安装执行brew install tesseract即可，安装过程可能有点慢，要下载很久
安装完成后，执行brew info tesseract，可以查看tesseract相关信息
使用
1 | tesseract <图片名字> <输出的文本文件名(自带.txt)> [-l 语言类型] [--oem ocrenginemode] [--psm pagesegmode] [configfiles...] |
PaddleOCR
识别率高
安装
pyCharm拉取github PaddlePaddle/PaddleOCR仓库的代码，并切换到分支
release/2.7在pyCharm终端执行下面命令,注意需要venv加载成功
1
2
3
4
5
6
7#环境所需依赖
(venv) ➜ PaddleOCR git:(release/2.7) pip install paddlepaddle
(venv) ➜ PaddleOCR git:(release/2.7) pip install -U pip setuptools
#安装
(venv) ➜ PaddleOCR git:(release/2.7) pip install "paddleocr>=2.0.1"
#使用
(venv) ➜ PaddleOCR git:(release/2.7) paddleocr --image_dir /Users/xuanleung/Downloads/ppocr_img/imgs/11.jpg --use_gpu false
实战场景
有一大批照片，要从照片中获取指定信息作为图片的文件名，并根据文件名生产csv表格
提前训练好模型参考：PaddleSeg基于机器学习的图像分割
采用paddleSeg（模型预测）进行关键位置的识别
1
2#pycharm 运行
python tools/predict.py --config configs/quick_start/pp_liteseg_optic_disc_512x512_1k.yml --model_path output/iter_1000/model.pdparams --image_path /Users/x/Downloads/桩顶标高 --save_dir /Users/x/Downloads/桩顶标高out --custom_color 0 0 0 255 255 255裁剪图片
1
2
3#pycharm 运行
python tools/crop.py --input_mask_dir /Users/x/Downloads/桩顶标高out/pseudo_color_prediction --input_orig_dir /Users/x/Downloads/桩顶标高
Creating crop directory: /Users/x/Downloads/桩顶标高/crop_dir执行ocr识别，会产生一个output目录，里面会产生和图片对应文件名的txt文本，内容为ocr识别的内容
1
2
3#建议优化脚本里面的识别方式，替换成PaddleOCR识别率提高非常的多
#在图片所在目录
bash ocr.sh执行关键文字提取脚本，将关键字在文本文件里面保留
1
2#在图片所在的output目录里面
bash format.sh替换文件名，根据文本里面的内容作为图片的文件名
1
2#在图片所在目录
bash rename.sh根据文件名生产csv表格
1
2#在图片所在目录
bash csv.sh
文件附录
ocr.sh：批量将当前脚本所在目录的图片转换成文本1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20#建议优化脚本里面的识别方式，替换成PaddleOCR识别率提高非常的多
#!/bin/bash
# 当前脚本所在目录
script_dir="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
# 输入文件夹路径为当前目录
input_folder="$script_dir"
# 输出文件夹路径为当前目录
output_folder="$script_dir/output"
# 创建输出文件夹
mkdir -p "$output_folder"
# 遍历输入文件夹中的所有图片文件
for image_file in "$input_folder"/*.jpg "$input_folder"/*.png "$input_folder"/*.jpeg; do
# 提取文件名（不含路径和扩展名）
filename=$(basename "$image_file" | cut -f 1 -d '.')
# 构建输出文本文件的路径
output_text_path="$output_folder/$filename"
# 执行 Tesseract 命令
tesseract "$image_file" "$output_text_path" -l chi_sim+equ --psm 6
echo "Processed $image_file"
doneformat.sh：批量将当前脚本所在目录的文本只保留关键信息1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17#!/bin/bash
# 获取当前脚本所在目录
script_dir="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
# 遍历当前目录下的所有文本文件
for file in "$script_dir"/*.txt; do
# 检查文件是否存在
if [ -f "$file" ]; then
# 使用grep命令查找匹配行并提取信息
transaction=$(grep -oE '体验[0-9]+' "$file" | cut -d'验' -f2)
elevation=$(grep -oE '[0-9]{3}\.[0-9]{3}' "$file" | cut -d':' -f2)
# 输出提取的信息
echo "交易编号: $transaction"
echo "高程: $elevation"
# 将提取的信息写入原文件
echo "$transaction-$elevation" > "$file"
fi
donerename.sh：批量重命名脚本所在目录的图片，文件名根据output目录对应文件名的文本内容进行命名1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30#!/bin/bash
# 获取当前脚本所在目录
script_dir="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
# 图片文件夹路径
image_folder="$script_dir"
# 遍历图片文件夹中的所有文本文件
for text_file in "$image_folder"/output/*.txt; do
# 检查文件是否存在
if [ -f "$text_file" ]; then
# 获取文本文件名（不含路径和扩展名）
base_name=$(basename "$text_file" .txt)
# 获取图片文件名
image_file="$image_folder/$base_name.jpg"
# 检查图片文件是否存在
if [ -f "$image_file" ]; then
# 获取文本文件的内容作为新的图片文件名
new_name=$(cat "$text_file")
# 检查是否有同名文件，如果有就给新名称添加序号
new_image_file="$image_folder/$new_name.jpg"
counter=1
while [ -f "$new_image_file" ]; do
new_image_file="$image_folder/${new_name}_${counter}.jpg"
((counter++))
done
# 重命名图片文件
mv "$image_file" "$new_image_file"
echo "Renamed '$image_file' to '$new_image_file'"
fi
fi
donecsv.sh：根据文件名生产表格，表格数据有两列，文件名以-进行区分列1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17#!/bin/bash
# 获取当前脚本所在目录
script_dir="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
# 图片文件夹路径
folder="$script_dir"
# CSV 文件路径
csv_file="$script_dir/output.csv"
# 创建 CSV 文件，并写入标题行
echo "Column1,Column2" > "$csv_file"
# 遍历文件夹中的所有图片文件
for file in "$folder"/*.jpg; do
# 提取文件名中的两个部分，使用 '-' 作为分隔符
part1=$(basename "$file" | cut -d'-' -f1)
part2=$(basename "$file" | cut -d'-' -f2 | sed 's/\.jpg//')
# 将两个部分写入 CSV 文件
echo "$part1,$part2" >> "$csv_file"
donecrop.py：根据模型预测的数据，进行图片裁剪1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45# 参数：input_mask_dir 掩码图片目录，黑白图片
# 参数：input_orig_dir 原始图片目录，文件名要和掩码图片文件名一致，后缀可以不一样
# 输出：根据掩码图片裁剪原始图片，得到原图的裁剪图片
import argparse
import glob
import os
import os.path as osp
import cv2
from pathlib import Path
from paddleseg.utils import logger
def parse_args():
parser = argparse.ArgumentParser(description='image crop.')
parser.add_argument('--input_mask_dir', help='input annotated directory',type=str)
parser.add_argument('--input_orig_dir', help='input annotated directory',type=str)
return parser.parse_args()
def main(args):
output_dir = osp.join(args.input_orig_dir, 'crop_dir')
if not osp.exists(output_dir):
os.makedirs(output_dir)
print('Creating crop directory:', output_dir) # 在原图目录创建裁剪后输出目录
for mask_img in glob.glob(osp.join(args.input_mask_dir, '*.png')): # 遍历input_mask_dir目录所有png的图片
print('read mask img from:', mask_img)
mask_img_name = Path(mask_img).stem # 获取文件名
print('read orig img name:', args.input_orig_dir+mask_img_name+".jpg")
# 读取原始图像和关键部位掩码图像
original_image = cv2.imread(args.input_orig_dir+"/"+mask_img_name+".jpg")
mask_image = cv2.imread(mask_img, cv2.IMREAD_GRAYSCALE) # 假设是单通道的灰度图像
# 查找掩码图像中的边界
contours, _ = cv2.findContours(mask_image, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
# 找到最大的边界
max_contour = max(contours, key=cv2.contourArea)
# 获取最大边界的边界框
x, y, w, h = cv2.boundingRect(max_contour)
# 从原始图像中裁剪关键部位
key_part = original_image[y:y + h, x:x + w]
# 保存裁剪出的关键部位
cv2.imwrite(output_dir+"/"+mask_img_name+".jpg", key_part)
logger.info(f'crop img is saved in {output_dir}')
if __name__ == '__main__':
args = parse_args()
main(args)
Docker安装升讯威在线客服与营销系统
安装
下载私有化部署包
解压下载的压缩包
在解压的目录新建一个
Dockerfile1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17FROM centos:centos7.9.2009
WORKDIR /wwwroot
COPY init.sh /tmp/sh/
COPY appsettings.json /tmp/config/
COPY Management /wwwroot/Management
COPY Resource /wwwroot/Resource
COPY Server /wwwroot/Server
RUN rpm -Uvh https://packages.microsoft.com/config/rhel/7/packages-microsoft-prod.rpm && yum -y install dnf && dnf install dotnet-sdk-3.1 -y && rpm -ivh https://dl.fedoraproject.org/pub/epel/epel-release-latest-7.noarch.rpm && yum install libgdiplus-devel nginx -y && ln -s /usr/lib64/libgdiplus.so /usr/lib/gdiplus.dll && ln -s /usr/lib64/libgdiplus.so /usr/lib64/gdiplus.dll && chmod +x /tmp/sh/init.sh && systemctl enable nginx
COPY default.conf /etc/nginx/conf.d/
EXPOSE 80
ENTRYPOINT ["/tmp/sh/init.sh"]创建初始化启动脚本,在解压的目录新建一个
init.sh1
2
3
4
5
6#!/bin/bash
cp /tmp/config/appsettings.json /wwwroot/Server/appsettings.json
nginx
nginx -s reload
cd /wwwroot/Server
dotnet Sheng.Linkup.Server.dllnginx代理配置,在解压的目录新建一个
default.conf1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72# https://kf.shengxunwei.com
# Start
map $http_upgrade $connection_upgrade {
default upgrade;
'' close;
}
upstream dotnet_server_proxy {
server localhost:5000;
keepalive 2000;
}
# kf-api (Server)
server{
listen 80;
listen [::]:80;
server_name kf-api.domain.io;
location / {
proxy_pass http://dotnet_server_proxy;
proxy_http_version 1.1;
proxy_set_header Upgrade $http_upgrade;
proxy_set_header Connection keep-alive;
proxy_set_header Host $host;
proxy_cache_bypass $http_upgrade;
proxy_set_header X-Forwarded-For $remote_addr;
proxy_set_header Upgrade $http_upgrade;
proxy_set_header Connection $connection_upgrade;
}
}
# kf-resource (Resource)
server {
listen 80;
server_name kf-resource.domain.io;
location / {
root /wwwroot/Resource;
index v.html;
}
error_page 500 502 503 504 /50x.html;
location = /50x.html {
root html;
}
}
# kf-m (Management)
server {
listen 80;
server_name kf-m.domain.io;
location / {
root /wwwroot/Management;
index index.html;
}
error_page 500 502 503 504 /50x.html;
location = /50x.html {
root html;
}
}
# End
# https://kf.shengxunwei.com在解压的目录新建一个
appsettings.json配置，里面进行mysql数据库配置，初始化脚本在：数据库建表脚本/CreateDatabase_MySql.sql1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54{
"Version": 3,
"Logging": {
"LogLevel": {
"Default": "Information",
"Microsoft": "Warning",
"Microsoft.Hosting.Lifetime": "Information"
}
},
"AllowedHosts": "*",
"DatabaseEngine": "mysql",
"ConnectionStrings": {
"DefaultConnection": "Server=172.1.1.2;database=kf_shengxunwei;user=root;password=Tech@2023"
},
"AppSettings": {
"FileStore": {
"Store": "static",
"AliOSS": {
"AccessKeyId": null,
"AccessKeySecret": null,
"BucketName": null,
"PublicAddress": null,
"EndPoint": null
}
},
"Email": {
"Account": "",
"Password": ""
},
"Translation": {
"Engine": "",
"Youdao": {
"AppKey": null,
"AppSecret": null
},
"Baidu": {
"AppId": null,
"AppSecret": null
}
},
"Environment": {
"DefaultLanguage": "zh-CHS",
"CustomerContext": {
"HeartbeatTimeoutSecond": 70,
"CellphoneHeartbeatTimeoutSecond": 70,
"OfflineOverSecond": 70
},
"TcpIpAddress": "172.1.1.44",
"TcpPort": "9527",
"ResourceAddress": "http://kf-resource.domain.com.cn",
"HostAddress": "http://kf-api.domian.com.cn"
}
}
}执行
docker build -t harbor.domain.dev/base/shengxunwei:1.3 .运行
docker run harbor.domain.dev/base/shengxunwei:1.3
配置
Resource对应客户访问的界面
Management对应客服管理后台界面
ApiUrl：http://kf-api.domain.io
ResourceUrl：http://kf-resource.domain.io
管理后台地址：http://kf-m.domain.io
需要修改下面几个文件里面的ApiUrl和ResourceUrl
1 | /wwwroot/Resource/embedded.js |
使用
客户访问http://kf-resource.domain.io/WebChat/MobileWebChat.html?siteCode=freesite
客服下载客户端，两个就能进行聊天了