Need to find perspective lines from reference image

Hello, I need to find vanishing lines or perspective lines from 2D images(Reference images). I tried many approaches like Line Segment Detector(LSD), RANSAC, HoughTransform, OpenCV, and some libraries. Still, they work well on Urban scenes and lane detection. But in my case, I have objects like almirahs, sofas, and showcases that detect wrong perspective lines and vanishing points. Below are the reference images I have -:

And below is the output I am receiving-:

And in the above images gives me the wrong results. Can someone guide me or share some resources? Is there any accurate method for finding the perspective lines and vanishing points? Thanks for your time.

Answer 1

[Sarthak]

Hi Gundeep,

As per my understanding, if existing approaches like LSD, RANSAC, etc are not working well for your images, you may need to fine-tune your parameters.

However, if you need accurate results you may need to consider using more advanced techniques like CNN to handle your specific cases.

Here are some resources for your reference.

https://www.mathworks.com/products/deep-learning.html

https://www.mathworks.com/help/vision/ug/object-detection-using-deep-learning.html