Pelee-TensorRT

Accelerate Pelee with tensorRT Pelee: A Real-Time Object Detection System on Mobile Devices (NeurIPS 2018)

TensorRT-Pelee can run over 200FPS on Titan V and 70FPS on jetson TX2(FP32)

Performance: 72 FPS, 13.8 ms (FP32) GPU Time:

Time by layers:

Requierments:

1.TensorRT4

2.CUDA 9.0

3.cudnn7

4.Jetpack 3.3(on TX2)

Run:

cmake .
make
./build/bin/pelee

Reference:

https://github.com/Ghustwb/MobileNet-SSD-TensorRT

TODO:

The bug has been fixed

Name		Name	Last commit message	Last commit date
Latest commit History 16 Commits
CMakeFiles		CMakeFiles
build		build
model/pelee		model/pelee
testPic		testPic
testVideo		testVideo
util		util
.DS_Store		.DS_Store
.gitignore		.gitignore
CMakeCache.txt		CMakeCache.txt
CMakeLists.txt		CMakeLists.txt
Makefile		Makefile
README.md		README.md
cmake_install.cmake		cmake_install.cmake
common.cpp		common.cpp
common.h		common.h
cudaUtility.h		cudaUtility.h
imageBuffer.h		imageBuffer.h
kernel.cu		kernel.cu
main.cpp		main.cpp
mathFunctions.cpp		mathFunctions.cpp
mathFunctions.cu		mathFunctions.cu
mathFunctions.h		mathFunctions.h
pluginImplement.cpp		pluginImplement.cpp
pluginImplement.h		pluginImplement.h
tensorNet.cpp		tensorNet.cpp
tensorNet.h		tensorNet.h

Provide feedback