This project shows how to run video decoding and object detection using simple pretrained Caffe model, yolov2
It is best to start doing these exercises after going through the presentation slides and MIVisionX github pages discussed earlier during this tutorial session. This exercise makes use for mv_compile utility which is built and installed with MIVisionX github repository https://github.com/GPUOpen-ProfessionalCompute-Libraries/MIVisionX.git. The tutorial exercise has two .cpp files, mvobj_detect.cpp and visualize.cpp. But it needs extra header files, .cpp files, and inference deployment library which is generated by mv_compile utility to completely build and execute the application.
For doing this exercise, follow the steps in order. For the second part of exercise, I will be showing how to run the object detection using 4 video streams by compiling model for batch of 4 and specifying 4 video files for the input to exercise 4.
- Ubuntu
16.04/18.04or CentOS7.5/7.6 - ROCm supported hardware
- AMD Radeon GPU or APU required
- ROCm
- Build & Install MIVisionX
- MIVisionX installs model compiler at
/opt/rocm/mivisionx - mv_compile installs at at
/opt/rocm/mivisionx/binand mvdeploy_api.h installs at/opt/rocm/mivisionx/include
- MIVisionX installs model compiler at
Step 2. Download pre-trained YoloV2 caffe model - yoloV2Tiny20.caffemodel
The mv_compile utility generates deployment library, header files, and .cpp files required to run inference for the specified model.
mv_compile --model yoloV2Tiny20.caffemodel --install_folder example4 --input_dims 1,3,416,416
There will be a file libmv_deploy.so (under ./lib), weights.bin and mvtestdeploy sample app (under ./bin). Also there will be mv_extras folder for extra post-processing helper functions. Open mvdeploy_api.h to go through API functions supported for inference deployment.
mvtestdeploy is a pregenerated application built during Step 2 which shows how to deploy inference for an input image file
cd example4
./bin/mvtestdeploy <inputdatafile> <output.bin> --install_folder . --t N
This runs inference for an input file and generate output for N number of iterations.
mv_objdetect is supposed to build on top of all the files generated in step 3. Basically it shows how to add preprocessing OpenVX nodes for video decoding and image_to_tensor conversion. Go through mv_objdetect.cpp file. This exercise uses a single or multiple video streams for input. The second part of the sample will show how to run it through multiple video files.
copy all files in clones sample folder (mvobjdetect.spp, visualize.cpp, visualize.h and CMakeLists.txt) into exercise4 folder everything is there to build and run the sample
cp mvobjdetect.cpp visualize.cpp .
cp visualize.h .
cp CMakeLists.txt .
mkdir build && cd build && cmake -DUSE_POSTPROC=ON ../
make -j
Note: if build directory exists from previous build, please remove it before creating again
./build/mvobjdetect ../data/img.jpg - --install_folder . --bb 20 0.2 0.4 --v
./build/mvobjdetect ../data/test.mp4 - --install_folder . --frames 5000 --bb 20 0.2 0.4 --v
Go thorugh steps 2 to 5, this time compiing the model for a batch of 4
mv_compile.exe --model yoloV2Tiny20.caffemodel --install_folder example4_batch4 --input_dims 1,3,416,416
cd example4_batch4
cp ../*.cpp .
cp ../*.h .
cp ../CMakeLists.txt .
mkdir build && cd build && cmake -DUSE_POSTPROC=ON ../
make -j
./build/mv_objdetect ../data/Videos_4.txt - --install_folder . --frames 5000 --bb 20 0.2 0.4 --v
This project is licensed under the MIT License - see the LICENSE.md file for details