mirror/opencv
1
0
Fork 0
mirror of https://github.com/opencv/opencv.git synced 2025-06-11 03:33:28 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

Updated Graph API (markdown)

Dmitry Matveev 2023-04-28 20:05:31 +01:00
parent 1e64483ea2
commit 43dc4fbeee
1 changed files with 1 additions and 46 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

47
Graph-API.md

@ -13,7 +13,7 @@ What it is?
* G-API GPU backend implements the majority of available functions and allows to run OpenCL kernels on available OpenCL-programmable devices. At the moment, GPU backend is based on OpenCV Transparent API; in future versions it will be extended to support integration of arbitrary OpenCL kernels (and likely be renamed to "OpenCL backend").
* G-API ONNX backend implements [ONNX models](https://github.com/onnx/models) inference operations on input data and outputs the results. At the moment, ONNX backend is based on [ONNX Runtime](https://github.com/microsoft/onnxruntime) C/C++ API.
* G-API ONNX backend implements [ONNX models](https://github.com/onnx/models) inference operations on input data and outputs the results. At the moment, ONNX backend is based on [ONNX Runtime](https://github.com/microsoft/onnxruntime) C/C++ API, build & run instruction can be found [here](https://github.com/opencv/opencv/wiki/Using-G-API-with-MS-ONNX-Runtime).
# Building G-API
@ -52,27 +52,6 @@ cmake /path-to-opencv -DPlaidML2_DIR=path-to-miniconda3/share/plaidml2 -DWITH_PL
/path-to-opencv-build/bin/opencv_test_gapi --gtest_filter=*GAPI_PlaidML_Pipelines*
```
## Building with Microsoft ONNX Runtime support
* Build and install the ONNX RT (currently tested with v1.14.1):
```bash
$ git clone --recursive https://github.com/microsoft/onnxruntime.git
$ cd onnxruntime
$ git checkout v1.14.1
$ git submodule update --init
$ ./build.sh --config Release --build_shared_lib --parallel \
$ --cmake_extra_defines CMAKE_INSTALL_PREFIX=/path-to-install-dir
$ cd build/Linux/Release
$ make install
```
* Then specify extra options to OpenCV CMake:
```bash
$ cmake /path-to-opencv -DWITH_ONNX=ON -DORT_INSTALL_DIR=/path-to-ort-install-dir
```
## Building with oneVPL Toolkit support
* Build (https://github.com/oneapi-src/oneVPL) or install `oneVPL` from installation manager (https://www.intel.com/content/www/us/en/developer/tools/oneapi/onevpl.html)
@ -230,30 +209,6 @@ $ ./downloader.py -o /path/to/opencv_extra/testdata/dnn/omz_intel_models/ \
--name age-gender-recognition-retail-0013
$ export OPENCV_DNN_TEST_DATA_PATH=/path/to/opencv_extra/testdata/dnn
```
## With ONNX Runtime
When you build G-API with ONNX Runtime support, tests for inference are enabled and require `OPENCV_GAPI_ONNX_MODEL_PATH` to be set:
```bash
$ export OPENCV_TEST_DATA_PATH=/path-to-opencv-extra/testdata
$ export OPENCV_GAPI_ONNX_MODEL_PATH=/path-to-onnx-models/
```
and [models](https://github.com/onnx/models) are downloaded using the commands:
```bash
$ git clone --recursive https://github.com/onnx/models.git
$ cd models
$ git lfs pull --include=path-to-desired-onnx-model --exclude=""
```
The following models are known to be required as for OpenCV 4.7:
```bash
git lfs pull --include vision/body_analysis/emotion_ferplus/model/emotion-ferplus-8.onnx --exclude=""
git lfs pull --include vision/classification/squeezenet/model/squeezenet1.0-9.onnx --exclude=""
git lfs pull --include vision/object_detection_segmentation/faster-rcnn/model/FasterRCNN-10.onnx --exclude=""
git lfs pull --include vision/object_detection_segmentation/ssd-mobilenetv1/model/ssd_mobilenet_v1_10.onnx --exclude=""
git lfs pull --include vision/object_detection_segmentation/tiny-yolov2/model/tinyyolov2-8.onnx --exclude=""
git lfs pull --include vision/object_detection_segmentation/yolov3/model/yolov3-10.onnx --exclude=""
```
Submitting G-API PRs
====================