n8886919 / yolo_alpr

Lincese Plate & OCR

Dependency: cuda10, ROS, conda

其實任何版本cuda都行,只是alpr.yml中的mxnet版本對應cuda10 Lincence Plate Detection權重連結 OCR權重連結

環境設置

sudo apt-get install ros-$version-usb-cam
sudo apt-get install ros-$version-cv-bridge
git clone https://github.com/n8886919/YOLO_ALPR
cd YOLO-ALPR
export PYTHONPATH=$PYTHONPATH:~/Desktop/YOLO:$(pwd)
conda env create -f alpr.yml
conda activate alpr.yml
cd licence_plate
python LP_detection.py v2 export --weight (Lincence_Plate_Detection_weight_unzip_path)
cd ../OCR
python OCR v1 export (OCR_weight_unzip_path)

Run Demo

車牌偵測

rosrun usb_cam usb_cam_node
py LP_detection.py v2 video (--topic your_video_topic)

車牌文字辨識

py OCR.py v1 video

Troubleshooting

if: :::danger [ERROR] [1552911199.414362]: bad callback: <bound method LicencePlateDetectioin._image_callback of <main.LicencePlateDetectioin instance at 0x7fd1aa0e1200>> Traceback (most recent call last): File "/opt/ros/melodic/lib/python2.7/dist-packages/rospy/topics.py", line 750, in _invoke_callback cb(msg) File "LP_detection.py", line 460, in _image_callback self.img = self.bridge.imgmsg_to_cv2(img, "bgr8") File "/opt/ros/melodic/lib/python2.7/dist-packages/cv_bridge/core.py", line 163, in imgmsg_to_cv2 dtype, n_channels = self.encoding_to_dtype_with_channels(img_msg.encoding) File "/opt/ros/melodic/lib/python2.7/dist-packages/cv_bridge/core.py", line 99, in encoding_to_dtype_with_channels return self.cvtype2_to_dtype_with_channels(self.encoding_to_cvtype2(encoding)) File "/opt/ros/melodic/lib/python2.7/dist-packages/cv_bridge/core.py", line 91, in encoding_to_cvtype2 from cv_bridge.boost.cv_bridge_boost import getCvType ImportError: /usr/lib/x86_64-linux-gnu/libblas.so.3: undefined symbol: sgemm_thread_nn ::: try:

sudo apt-get remove libopenblas-base