TCT模型部署步骤(3:代码修改)

最新推荐文章于 2024-09-27 21:44:24 发布

南城同学

最新推荐文章于 2024-09-27 21:44:24 发布

阅读量1.6k

点赞数

分类专栏：模型部署文章标签：机器学习

本文链接：https://blog.csdn.net/Nine_Yao/article/details/122958425

版权

模型部署专栏收录该内容

6 篇文章 0 订阅

订阅专栏

1. 代码层面

TctDetect项目，main.cpp下:


        //本机是单显卡，这里改为"false"
24     const bool TwoDetect = false;	

---
       //因为是单显卡，所以把79、80、82、83、84行注销掉
78		ObjectDetect od1(&pp, 1, inferConf.posEnginePath, inferConf.negEnginePath, inferConf.gpuid, inferConf.tileSize, inferConf.modelSize, size[0], size[1], inferConf.iouThres, inferConf.posConfThres, inferConf.negConfThres, inferConf.batchSize, inferConf.maxBatchNum, 1);
79		//ObjectDetect od2(&pp, 2, inferConf.posEnginePath, inferConf.negEnginePath, inferConf.gpuid2, inferConf.tileSize, inferConf.modelSize, size[0], size[1], inferConf.iouThres, inferConf.posConfThres, inferConf.negConfThres, inferConf.batchSize, inferConf.maxBatchNum, 1);
80		//vector<ObjectDetect*> ods = { &od1, &od2 };
81		vector<ObjectDetect*> ods = { &od1 };
82		//if (TwoDetect) {
83		//	ObjectDetect od2(&pp, 2, inferConf.posEnginePath, inferConf.negEnginePath, inferConf.gpuid2, inferConf.tileSize, inferConf.modelSize, size[0], size[1], inferConf.iouThres, inferConf.posConfThres, inferConf.negConfThres, inferConf.batchSize, inferConf.maxBatchNum, 2);
84		//	ods.push_back(&od2);
85		//}

2. Visual Studio配置修改

2.1. TctDetect项目

设为"启动项"；
属性 --> 链接器 --> 命令行，在原内容后面空格添加/FORCE:MULTIPLE ；

 %(AdditionalOptions) /machine:x64 /FORCE:MULTIPLE

属性--> 调试 --> 环境，添加:

PATH=E:\Dev\TCTDetect\libs\TensorRT-7.2.3.4\lib;E:\Dev\TCTDetect\libs\opencv452\build\x64\vc15\bin;E:\Dev\TCTDetect\libs\sdpc\lib\WINDOWS\x64\Release\dll;E:\Dev\TCTDetect\libs\openslide-win64-20171122\bin;C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.1\bin;