This folder contains code (packaged into a Docker Container) to auto-label a single image using groundingDino
To run locally:
- Go to handler.py and uncomment the section to write data to a json file (RunPod expects a 'test_input.json' file when running the script locally)
- Navigate to the terminal and run
cd inference/RunPod&docker-compose up
To deploy to RunPod: https://www.notion.so/tapwayops/Deploy-groundingDino-onto-RunPod-1dfb42ac18914222a99b71c278248de5?pvs=4
To test serverless deployment:
- Create a
.envfile with the env variableRUNPOD_API_KEY=<your_RunPod_api_key> - Navigate to
test_single.pyto autolabel 1 image ortest_multi.pyto autolabel many images - (optional) Put your own image(s) into the folder under
data/samples/(if usingtest_single.py, make sure to change theimgvariable to your image filename) - Navigate to the terminal and run
cd inference/RunPod&python3 test_single.pyorpython3 test_multi.py - View your images at
data/output/
General Info on RunPod: https://www.notion.so/tapwayops/Research-into-serverless-deployment-providers-31a9caa0875247fc870a68ad9cb9c706?pvs=4