Earn 18,000 ($180.00)

What is the Goal ?
Given an image get segments of a room - Wall, furniture, flooring(rug, carpet, wood) & wall-art-work. These segments can then be analyzed. In the first phase I just need dominant color in each image segment. Second phase will expand on these functions.

How will it work for the user ?

• User drops an image & presses a button called "Analyze".
• The program analyzes the image & creates 4 segments - (Walls, Furniture, Flooring & Wall-Art-work)
• In a small table at the bottom of the page, shows the dominant color of each
• If program fails at analyzing it asks user to click on each segment for each category step by step - e.g. "Click on all segments which are wall"

Technology

• Either use SAM(Segment Anything by Meta - https://segment-anything.com/)
• Or if there is any other segmentation commercially available model - only if SAM isn't viable
• Here is one implementation https://github.com/facebookresearch/segment-anything/blob/main/notebooks/predictor_example.ipynb
• If easier, create a replit that has full functionality before you deploy to AWS
• I'm ok if you want to use ChatGPT to analyze the segmented items.

Note
Please don't apply if

• You can't speak on the phone with me
• If you don't have experience in image segmentation

Screening Questions

• Give me your honest opinion on feasibility of project
• What are your thoughts on the model to use if it's not SAM
• Tell me about your image segmentation experience
• Tell me if you've used SAM