0
$\begingroup$

I have an image of a one-line substation schema diagram that includes various components (like transformers, circuit breakers, etc.) and the connections between them. I’m looking for a way to convert this image into a structured format (XML or JSON) that accurately captures the following details:

  • Component Details: The exact coordinates and types of each element detected in the image.
  • Connections: All the interconnections (lines) between the components.
  • Reconstruction: The resulting XML/JSON should be sufficient to later reconstruct the original diagram.

So far, I’ve experimented with object detection models such as YOLO, but the results aren’t detailed enough, especially regarding the connections between components.

My questions are:

  1. Image Processing Techniques: What methods can be used to accurately detect both the individual components and the connecting lines? Are there any specific computer vision techniques or libraries (e.g., edge detection, Hough transform) that could help with this task?
  2. Deep Learning Approaches: Which deep learning models (like Mask R-CNN, DETR, etc.) or hybrid approaches could be effective for this problem, especially when pre-trained models don’t yield the necessary precision?

Substation one line schema

$\endgroup$
3
  • $\begingroup$ context: stackoverflow.com/staging-ground/79494493 -- also there: ai.stackexchange.com/questions/48194/… $\endgroup$ Commented Mar 9 at 11:58
  • $\begingroup$ Not sure this is the best site to ask that $\endgroup$ Commented Mar 14 at 8:23
  • $\begingroup$ @LucasMorin someone in the first question's staging ground listed a bunch of other SEs with vague relation to the question, in no particular order, so that's how this happened. that was a high-rep user with no expertise in the subject matters required for this question. I think this question is too broad for any site, unless it accepts opinions/discussions. $\endgroup$ Commented Mar 14 at 11:40

0

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.