RoboflowOak / Oak-D Pro W Camera / Offline Usage

Duane

Hello,

I'm working with Roboflow and a trained model. I've successfully got a raspberry pi to run the Oak-D W Camera and do detection based on the model using the RoboflowOak example below:
https://blog.roboflow.com/opencv-ai-kit-deployment/

However our usages it to have the robot NOT connected to the Internet. So when we bridge the robot to the internet all works great. When we disconnect it the camera code crashes with the below error:

urllib3.exceptions.MaxRetryError: HTTPSConnectionPool(host='api.roboflow.com', port=443): Max retries exceeded with url: /oak/crescendonote/1?api_key=[Redacted Key]&device=19443010E15EA12E00&nocache=true (Caused by NewConnectionError('<urllib3.connection.HTTPSConnection object at 0x7fbd125710>: Failed to establish a new connection: [Errno -3] Temporary failure in name resolution'))

Is it not possible to use these cameras with a cached model and use the roboflowoak library? is there an easier tutorial you could point me at?

Thanks,

Duane

jakaskerl

Hi @Duane
Not sure about the roboflowoak API, but if you use the SDK: https://docs.luxonis.com/projects/sdk/en/latest/samples/NNComponent/sdk_roboflow_integration/ the model should get stored in ~/.cache/roboflow-models/.

Thanks,
Jaka

Duane

This has helped out a TON thank you.

Duane

Fixed a small issue with the code in the integrations/roboflow.py file. The integration was making a request to api.roboflow.com prior to checking to see if the cache existed. This caused offline systems to throw an exception. I've put in a pull request here luxonis/depthai1154 to merge that change back into the main repo.

Thanks again for pointing me at the depthai_sdk!

jakaskerl

Duane
Awesome, thanks.