• 状态 Resolved
  • 已锁定 已锁定
  • 个回复 4 个回复
  • 名订阅用户 15 名订阅用户
  • 次查看 548 次查看
  • 用户 0 名在线用户
选项
选项
相关

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

SK-AM68: Is able to run object detection from command line with SK-AM68 ?

greatcat chen
Prodigy 50 points
Part Number: SK-AM68

Hi, Experts:

I bought SK-AM68 and tried demo of "Edge AI gallery".
I noticed the left side function of "Custom". The function is able to select Input type, Image and Model with GUI.
How to execute ONR-OD-xxx or TVM-OD-xxx from the command line?
In the case, is able to return resoult by file ?
Or, is there any API able to call for this?

By the way, I want to let SK-AM68 as a co-processor for object detection.
I think that is the easy/simple/lazy/fastest way to test/demo/apply/deploy TI AI solution.
Using gigaLAN/TCP input JPG and output/return result by XML/JSON/text.
If I finish this I will publish the code to gitHub.

Best Regards,

Greatcat

  • 0
    TI__Mastermind 27060 points

    Hello，

    greatcat chen 说：
    I bought SK-AM68 and tried demo of "Edge AI gallery".
    I noticed the left side function of "Custom". The function is able to select Input type, Image and Model with GUI.
    How to execute ONR-OD-xxx or TVM-OD-xxx from the command line?
    In the case, is able to return resoult by file ?
    Or, is there any API able to call for this?

     I don't quite understand what you mean. Can you rephrase it?

    greatcat chen 说：
    If I finish this I will publish the code to gitHub.

    Wishing you all the best. Thank you for sharing

    Regards，

    Gary

  • 0 回复 Gary Lu
    Prodigy 50 points

    Hi, Gary:

    For example,

    input as :

    http://greatcat.url.tw/q1.jpg

    results as:

    http://greatcat.url.tw/q2.jpg

    Can I run ONR-OD-8220-yolox-s-lite-mmdet-coco-640x640 by minicom/UART ?

    May I get the results as below in file?

    (1)the 2 of strings : "vase" "food/orange"

    (2)the 2 of green rectangle x-y position and width / height

    Best regards,

    Greatcat

  • 0 回复 greatcat chen
    TI__Mastermind 27060 points

    Hello Greatcat，

    To run the ONR-OD-8220-yolox-s-lite-mmdet-coco-640x640 model and obtain the results through a command-line interface, you'll typically need a software stack that includes the necessary tools and libraries for running inference with this specific model.

    For instance, if you're using a Python script to run inference, you can use libraries like OpenCV or TensorFlow to load the image and process the results. Then, you can use file I/O operations in Python to write the results to a file.

    Regards，

    Gary

  • 0 回复 Gary Lu
    Prodigy 50 points

    Hi, Gary:

    I see that the OS comes with official command line demo at 

    /opt/edgeai-gst-apps/tests/

    Test step:

    (1)

    /etc/init.d/edgeai-launcher.sh stop

    (2)

    cd /opt/edgeai-gst-apps/tests/

    (3)

    sh test_all_models.sh

    The demo supports python and CPP. 

    The config file /opt/edgeai-gst-apps/tests/test_config.yaml comes with some magic parameter.

    I extract them and put to 

    https://github.com/greatcattw/TI_SK68_edgeai_yaml

    Then test by command line (python)
    /opt/edgeai-gst-apps/apps_python/app_edgeai.py -n -v /opt/edgeai-gst-apps/tests/t1/ONR-OD-8420.yaml
    or (CPP)
    /opt/edgeai-gst-apps/apps_cpp/bin/Release/app_edgeai -n -v /opt/edgeai-gst-apps/tests/t1/ONR-OD-8420.yaml

    Best regards,

    Greatcat