This demo will show you how to use libtorch to build your C++ application.
- Pytorch (tag: pytorch v1.0)
- Libtorch
- OpenCV
Export your pytorch model to torch script file, We will simply use resnet50 in this demo
Write your C++ program, check the file prediction.cpp for more detial.
PS: module->to(at::kCUDA) and input_tensor.to(at::kCUDA) will switch your model & tensor to GPU mode,
comment out them if you just want to use CPU mode.
Write a CMakeLists.txt, the version of OpenCV must the same as your libtorch.
Otherwise, you may get the compile error:
error: undefined reference to `cv::imread(std::string const&, int)'
check issues 14684 and issues 14620 for more details.
- run
model_trace.py, then you will get a fileresnet50.pt - compile your cpp program, you need to use
-DCMAKE_PREFIX_PATH=/absolute/path/to/libtorch, for example:
mkdir build
cd build
cmake -DCMAKE_PREFIX_PATH=/home/cgilab/pytorch/torch/lib/tmp_install ..
make
- test your program
classifier <path-to-exported-script-module> <path-to-lable-file>
> ./classifier ../resnet50.pt ../label.txt
== Switch to GPU mode
== ResNet50 loaded!
== Label loaded! Let's try it
== Input image path: [enter Q to exit]
../pic/dog.jpg
== image size: [976 x 549] ==
== simply resize: [224 x 224] ==
============= Top-1 =============
Label: beagle
With Probability: 99.1228%
============= Top-2 =============
Label: Walker hound, Walker foxhound
With Probability: 0.469356%
============= Top-3 =============
Label: English foxhound
With Probability: 0.110916%
== Input image path: [enter Q to exit]
../pic/shark.jpg
== image size: [800 x 500] ==
== simply resize: [224 x 224] ==
============= Top-1 =============
Label: tiger shark, Galeocerdo cuvieri
With Probability: 92.2599%
============= Top-2 =============
Label: great white shark, white shark, man-eater, man-eating shark
With Probability: 5.94252%
============= Top-3 =============
Label: hammerhead, hammerhead shark
With Probability: 1.77418%
== Input image path: [enter Q to exit]
Q
Take it easy!!