Help the visually impaired navigate your building
We start by extracting all straight edges in the floor plan image. Afterwards, we merge lines that are close proximity to one another. The result is what a human looking at the floor plan would concider to be walls.
After line identification, we identify potential square areas that a door might be located. Then, using feature detection, we isolate the image of door and use feature detection to identify the rest of the doors in the image.
The current text detection algorithm's performance is effective when texts are the only observable object in the image. We first isolate the text by identifying conjoined lines. We then use Tessearct to perform OCR on the isolated text.
Upload floor plans of public spaces that you have the rights to
Close off all rooms, hallways, staircases and other features
Select the portions of walls that are doorways, entrances or exits
Label closed off areas as either room, hallway, staircase or elevator.
Pin points of interest such as bathrooms, drinking fountains, etc.
Connect the floors of your building with staircases and elevators
The graph consists of three levels: at the top is floor connections, next is the floor graph and at the lowest level is the grid view. The first level represents how floors connect to each other. The second level consists of the floor graph which represents how rooms, passageways, hallways, stairs, and elevators connect to each other. The first two levels are used for high level path planning, which calculates the path of nodes between the source and destination. The third level is key in low level planning, which constructs a detailed path to aid the visually impaired avoid obstacles and find passageways.