I have an application, pretty much openCv code, and I don't know how to put this on pipeline to be processed in OAK device. Any help?

    Hi Lucas_Lima This is something I have implemented a few days ago, here - https://github.com/dhruvsheth-ai/gen2-depthai-experiments

    This repo contains examples of using OpenCV functionalities like Background Subtraction, Edge and Corner detection, Hough transform etc. But this is being processed on host, with data retireved from OAK-D. OpenCV functionalities are light, so have kept it on host, but will migrate to processing on OAK-D for a later version.

    Also, you'll find this example - https://github.com/luxonis/depthai-python/blob/main/examples/20_color_rotate_warp.py which uses pure OpenCV functions to do the task.

    There might be some more examples, I might not be aware of, but if there's anythin I can help here, let me know!
    Thanks

        dhruvsheth-ai I really appreciate your reply. I'd like to know how you intend to migrate processing from host to OAK using DAI library. I've read the documentation sometimes and I couldn't find how to create a custom node in pipeline with my own functions

            Lucas_Lima I'd like to know how you intend to migrate processing from host to OAK using DAI library.

            I'm not pretty sure at the moment, but I should get back to you within a week. The main reason I've kept it on host is that those functions do not use a lot of CPU.

                12 days later

                Lucas_Lima I don't believe it's possible (except for using the script node) since firmware is not open source. So best idea is to make the blob (as mentioned above) that would do it for you.
                Thanks, Erik