[TASK] Deep Learning Model for Valve Handle Angle Detection

[kluge7]

Description of task

Replace the current Hough-transform–based handle angle detection with a dedicated deep learning model. The model will take cropped valve images (from existing valve detection) and predict the handle’s rotation angle directly.

Suggested Workflow

No response

Specifications

• Input bounding boxes will come from the valve detection model and may vary in size. The angle detection node must handle variable input sizes by scaling, padding, or resizing crops before inference.
• Output should be publised as a ROS message containing: std_msgs/Float32 for the detected angle in degrees (0° = closed reference position, increasing counterclockwise).

Contacts

• @jorgenfj
• @kluge7

Code Quality

• Every function in header files are documented (inputs/returns/exceptions)
• The project has automated tests that cover MOST of the functions and branches in functions (pytest/gtest)
• The code is documented on the wiki (provide link)