YoloV5 ML backend for label-studio for semi-automatic labeling
Step-0: install label-studio-ml-backend using pip and if necessary then install label-studio dependencies.
pip install git+https://github.com/heartexlabs/label-studio-ml-backend.git
Step-1: Copy yolov5 directory to your label-studio project
Step-2: Change yolov5/model.py settings if you want to use custom settings. Change following setttings to work with custom model.
# COCO pretrained model
model = torch.hub.load('ultralytics/yolov5', 'yolov5s')
# Custom model
model = torch.hub.load('ultralytics/yolov5', 'custom', path='path to your model')
Do not forget to modify labelmap in model script. Change it to your model labelmaps.
Step-3: Initialize model inference server
label-studio-ml init yolov5_ml --script yolov5/model.py
Step-4: Start model inference server
label-studio-ml start ./yolov5_ml
Copy inference url for next stop. It will be shown in blue color i.e. http://192.168.0.10:9090/
Step-5: Open your label-studio user interface then add model to your settings by
settings > machine learning > Add model url with model name > save
Step-6: Enjoy Semi-Automatic labeling using YoloV5