gpu support for autolabeling is now possible for yolov5

anylabeling/services/auto_labeling/yolov5.py

@@ -38,6 +38,8 @@ def __init__(self, model_config, on_message) -> None:
             raise Exception(f"Model not found: {model_abs_path}")
 
         self.net = cv2.dnn.readNet(model_abs_path)
+        self.net.setPreferableBackend(cv2.dnn.DNN_BACKEND_CUDA)
+        self.net.setPreferableTarget(cv2.dnn.DNN_TARGET_CUDA)
         self.classes = self.config["classes"]
 
     def pre_process(self, input_image, net):

anylabeling/views/labeling/label_file.py

@@ -167,7 +167,7 @@ def save(
         if flags is None:
             flags = {}
         data = dict(
-            version=__version__,
+            #version=__version__,
             flags=flags,
             shapes=shapes,
             imagePath=image_path,

opencv_dnn.sh

@@ -0,0 +1,66 @@
+# Most of the steps were done following this: https://pyimagesearch.com/2020/02/03/how-to-use-opencvs-dnn-module-with-nvidia-gpus-cuda-and-cudnn/
+
+# Download openCV and create your environment:
+cd ~
+mkdir .opencv_drivers
+cd .opencv_drivers/
+wget -O opencv.zip https://github.com/opencv/opencv/archive/4.6.0.zip
+wget -O opencv_contrib.zip https://github.com/opencv/opencv_contrib/archive/4.6.0.zip
+unzip opencv.zip
+unzip opencv_contrib.zip
+mv opencv-4.6.0 opencv
+mv opencv_contrib-4.6.0 opencv_contrib
+mkvirtualenv opencv_cuda
+pip install numpy
+cd opencv
+mkdir build
+cd build
+
+# IMPORTANT!: CUDA_ARCH_BIN version must be seen here: https://developer.nvidia.com/cuda-gpus
+# I got RTX 2060, so in my case CUDA_ARCH_BIN=7.5
+# gcc version > 8 are not supported so set CMAKE_C_COMPILER=/usr/bin/gcc-8
+
+cmake -D CMAKE_BUILD_TYPE=RELEASE \
+	-D CMAKE_INSTALL_PREFIX=/usr/local \
+	-D INSTALL_PYTHON_EXAMPLES=ON \
+	-D INSTALL_C_EXAMPLES=OFF \
+	-D OPENCV_ENABLE_NONFREE=ON \
+	-D WITH_CUDA=ON \
+	-D WITH_CUDNN=ON \
+	-D OPENCV_DNN_CUDA=ON \
+	-D ENABLE_FAST_MATH=1 \
+	-D CUDA_FAST_MATH=1 \
+	-D CUDA_ARCH_BIN=7.5 \
+	-D WITH_CUBLAS=1 \
+	-D OPENCV_EXTRA_MODULES_PATH=~/.opencv_drivers/opencv_contrib/modules \
+	-D HAVE_opencv_python3=ON \		
+	-D PYTHON_EXECUTABLE=~/.virtualenvs/opencv_cuda/bin/python \
+	-D BUILD_EXAMPLES=ON \
+	-D CMAKE_C_COMPILER=/usr/bin/gcc-8 ..
+
+# Compile your openCV built before
+make -j12
+sudo make install
+sudo ldconfig
+
+
+# ln -s command is used to create a symbolic link, also known as a symlink or soft link
+cd ~/.virtualenvs/opencv_cuda/lib/python3.8/site-packages/
+ln -s /usr/local/lib/python3.8/site-packages/cv2/python-3.8/cv2.cpython-38-x86_64-linux-gnu.so cv2.so
+
+# Install all requirements for anylabeling
+pip install -r requirements.txt
+pip install -r requirements-dev.txt
+
+# Uninstall openCV dependencies, since dnn module installed from source is not recognize by pip
+pip uninstall opencv-contrib-python-headless opencv-python-headless
+
+
+
+
+
+
+
+
+
+

requirements.txt

@@ -18,4 +18,4 @@ PyYAML==6.0
 colorama==0.4.5
 onnx==1.13.1
 onnxruntime==1.14.1
-qimage2ndarray==1.10.0
+qimage2ndarray==1.10.0