Floor Plan Segmentation and Room Labeling (Roboflow)

About OpenTrain

OpenTrain is the #1 platform for finding and building careers in AI training and data labeling. We connect people with projects that teach AI how to see, read, and understand the world.

As an OpenTrain contributor you join a fast-growing industry where human examples and careful labeling directly shape how state-of-the-art AI systems behave.

• Create a free OpenTrain account to apply and build a profile.
• Work remotely, choose flexible hours, and gain experience in AI training and computer vision.

About AI training for computer vision

AI training (data labeling) is the human work behind machine perception. For computer vision, that means drawing precise masks and adding labels so models learn to detect and classify objects and regions.

This project focuses on architectural floor plans—an important real-world application where accurate segmentation and occupancy classification enable search, analytics, accessibility and downstream automation.

• You will help train models that detect rooms and assign occupancy types from scanned or rasterized architectural drawings.
• No prior AI research experience required—attention to detail and consistent labeling are key.

The role

We are hiring part-time contract labelers to find and annotate architectural floor-plan images using Roboflow. You will draw polygon masks for each room and tag them with occupancy types.

This is an intermediate-level labeling role that requires experience with polygon/segmentation tools and the ability to interpret basic architectural drawings. The work is batch-based and remote.

• Employment type: Contractor, Part-time.
• Time requirement: 20+ hours per week.
• Pay: USD $10.00 per hour.

What you'll do

Apply polygon and segmentation tools in Roboflow to create clean, non-overlapping masks and attach the correct room-type labels.

Verify that printed text labels and abbreviations in the plans match the assigned occupancy type and correct any mismatches according to the guidelines.

• Draw polygon masks around each room boundary so polygons follow printed walls and avoid overlaps.
• Label every room with an occupancy type (e.g., Bedroom, Bathroom, Corridor, Storage, Mechanical, Office).
• Review and confirm room-name text matches the selected occupancy tag.
• Complete datasets where each file typically contains 15–20+ labeled items (polygons + tags).

Dataset details

Source images are extracted from architectural PDFs and high-resolution raster images. Floor plans vary in style, clarity, and density—covering residential, commercial, and mixed-use buildings.

Detailed written instructions and examples will be provided for common symbols, abbreviations, and edge cases; maintain consistency across batches.

• Files include varied drawing styles and image quality—expect noise, differing label conventions, and complex room configurations.
• Each file requires thorough, accurate annotation to support segmentation model training.

Requirements

Candidates must have hands-on experience with polygon/segmentation labeling in Roboflow or similar tools and be comfortable reading basic architectural floor plans.

Strong attention to detail, consistent quality, and reliable communication are required for batch-based work and deadlines.

• Prior experience with Roboflow, CVAT, Labelbox, or similar polygon/segmentation tools.
• Ability to interpret walls, room boundaries, doors, text labels and common floor-plan abbreviations.
• Comfortable following detailed written instructions and checking label consistency.
• Reliable communication and ability to meet deadlines for assigned batches.
• Preferred (not required): background in architecture, drafting, civil engineering, or related fields.

Who should apply and how it works

This role is a great fit for intermediate annotators with polygon/segmentation experience who want steady part-time work contributing to a practical computer-vision dataset.

Apply through your OpenTrain profile, complete any onboarding tasks, review the labeling guide and examples, then accept batches and submit annotated files through Roboflow per project instructions.

• Worldwide applicants accepted; this is remote work.
• Onboarding includes a short qualification or sample task to ensure consistency with project standards.
• Work is paid hourly at the listed rate; contractors are responsible for their own taxes and compliance.