{When you're done, you can delete the content in this README and update the file with details for others getting started with your repository}

Software architecture description

bellow is the installation instruction for configuration of CUDA 11.1 + cudnn 8.0 + TensorRT 7.2.2.3 for cuda11.1 cudnn 8.0 + conda with python 3.8 (lj_torch)+ pytorch1.8.1 for cuda11.1. You can adopt other configurations and accordingly modify the version name in the following commands.

(1) CUDA 11.1 + cudnn 8.0 sudo ln -s cuda-11.1 cuda # for all users, do this only once export PATH=/usr/local/cuda-11.1/bin:$PATH export LD_LIBRARY_PATH=/usr/local/cuda-11.1/lib64:$LD_LIBRARY_PATH export CUDA_DEVICE=0 # may be needed for some codes

(2) TensorRT 7.2.2.3 for cuda11.1 cudnn 8.0

tar -xzvf TensorRT-7.2.2.3.Ubuntu-18.04.x86_64-gnu.cuda-11.1.cudnn8.0.tar.gz (for all users, do this only once) sudo cp -r TensorRT-7.2.2.3 /usr/local # for all users, do this only once export PATH=/usr/local/TensorRT-7.2.2.3/bin:$PATH export LD_LIBRARY_PATH=/usr/local/TensorRT-7.2.2.3/lib:$LD_LIBRARY_PATH

cd /usr/local/TensorRT-7.2.2.3/samples/trtexec sudo make clean & sudo make # for all users, do this only once

(3) conda environment(python3.8 + pytorch1.8.1 for cuda11.1 + TensorRT-7.2.2.3 for python) conda create -n lj_torch python=3.8 conda activate lj_torch pip3 install torch==1.8.1+cu111 torchvision==0.9.1+cu111 torchaudio==0.8.1 -f https://download.pytorch.org/whl/torch_stable.html cd /usr/local/TensorRT-7.2.2.3/python pip3 install tensorrt-7.2.2.3-cp38-none-linux_x86_64.whl # copy the .whl file to a folder needs no sudo cd /usr/local/TensorRT-7.2.2.3/graphsurgeon pip3 install graphsurgeon-0.4.5-py2.py3-none-any.whl # copy the .whl file to a folder needs no sudo cd /usr/local/TensorRT-7.2.2.3/onnx_graphsurgeon pip3 install onnx_graphsurgeon-0.2.6-py2.py3-none-any.whl # copy the .whl file to a folder needs no sudo pip3 install pycuda

1. (Optional) create an onnx model to test, run the follong code in python in your prepared conda enrionment: import torchvision.models as models resnext50_32x4d = models.resnext50_32x4d(pretrained=True) import torch BATCH_SIZE = 64 dummy_input=torch.randn(BATCH_SIZE, 3, 224, 224) import torch.onnx torch.onnx.export(resnext50_32x4d, dummy_input, "test.onnx", verbose=False)

2. (Optional) use trtexec to compare with this project, or to check if TensorRT installed correctly, e.g. [TensorRT install path]/trtexec --onnx=test.onnx --saveEngine=test.trt --explicitBatch --verbose

3. do the test and make sure no error appear python test_model_convertor.py (before running, modify paths in this script) python test_mmdet_ssd.py (before running, modify paths in this script)

4. (Optional) to test your other models, modify in load_test_model in file test_model_convertor.py

5. call ModelConvertor from your codes

1. Fork the repository
2. Create Feat_xxx branch
3. Commit your code
4. Create Pull Request

1. You can use Readme_XXX.md to support different languages, such as Readme_en.md, Readme_zh.md
2. Gitee blog blog.gitee.com
3. Explore open source project https://gitee.com/explore
4. The most valuable open source project GVP
5. The manual of Gitee https://gitee.com/help
6. The most popular members https://gitee.com/gitee-stars/