Help with Implementing Custom Object Detection in Mech-Eye SDK

Mech-Eye SDK Development
1

Hi everyone,

I am working on integrating custom object detection into the Mech–Eye SDK but running into some challenges. Specifically, I am trying to configure the SDK to recognize objects based on unique shapes and colors not predefined in the standard library.

Here’s what I’ve done so far ::-

  • Installed the latest version of the Mech–Eye SDK.
  • Followed the basic setup instructions for object detection.

However, I am struggling with the configuration settings and the code for training the model with my own data.
Could anyone share insights or point me to resources on customizing object detection with Mech-Eye SDK: ??
Any tips or examples would be greatly appreciated!!! I also read this official release: https://docs.mech-mind.net/en/eye-3d-camera/2.2.0/getting-started/release-notes.msbi but couldn’t get any help.

Thanks in advance… :grinning:

James

2

Hi James,

I appreciate you reaching out. However, it’s difficult for us to provide direct assistance with the issue you’ve raised for the following reasons:

  • The Mech–Eye SDK does not provide object detection interface; it only offers basic functions such as image capture, point cloud acquisition, and calibration. Reference: Mech-Mind · GitHub

    Through Mech-Eye SDK, you can control Mech-Eye Industrial 3D Camera (hereinafter referred to as “the camera”) to obtain images and point clouds. Available for Windows and Linux, Mech-Eye SDK includes Mech-Eye API (the native API) and the GenICam interface for the camera. Mech-Eye API is available for C++, C#, and Python. ROS 1 and ROS 2 interfaces are also provided for communicating with the camera.

  • We generally use Mech-Vision software for object detection, but these algorithms are encapsulated within the software.

Unfortunately, we cannot directly assist you,but, here are some references that might help you further:

Best regards,
Shawn