Hi everyone,
I’m working on a project where I need to extract the room layout from architectural floorplans. The goal is to process images of these floorplans and identify the boundaries of each room, differentiating between them accurately using their limiting walls.
Besides walls, that have different representations depending on their thickness and material, the images contain other elements such as:
- Doors
- Labels
- Measurements
- Furniture
- Diagrams
Here is an example of how they can look:
Despite these complexities, I want to create a model that can segment each separate room and provide an output recognizing distinct areas, each representing a different room.
I’ve researched a bit and believe this is primarily a segmentation task. However, I’m not sure about the best approach or models to use for this specific problem. Here are some of the questions I have:
- What types of models or algorithms are best suited for segmenting rooms in architectural floorplans?
- Are there any specific datasets or pre-trained models that could provide a good starting point for this task?
- How should I handle the diverse elements (like furniture and labels) that might interfere with the room boundaries?
- What are the best practices for annotating the data to train the model effectively?
Any guidance, resources, or examples of similar projects would be greatly appreciated!
Thanks in advance for your help!