Computer Vision Systems
Built an AI-powered system for pre-construction floor plan drawings - capable of detecting different objects with high accuracy. We also developed a system to detect the speed of the cars in real time for automated highway monitoring.
Year :
2025
Industry :
Tech
Problem :
Architects, contractors, and DIY teams waste time tracing walls from photos or floor plans; results vary by person, and tricky inputs—dim lighting, skewed angles, or faded scans—cause misses that ripple into bad measurements, wrong material estimates, and rework later.
Solution :
I have created a Streamlit app that turns wall extraction into a one‑click flow: upload an image or plan, preview an instant overlay of detected walls, tweak sensitivity if needed, and export the result for takeoffs or CAD handoff; it standardizes detection quality and makes the task fast for non‑experts, too.
Challenge :
Delivering robust detection across messy real‑world inputs—from hand‑drawn sketches to smartphone photos—while keeping latency low in the browser, and gracefully handling edge cases like overlapping lines, reflective surfaces, and low‑dpi scans without overwhelming users with controls.
Summary :
A practical wall‑detection assistant for pre‑construction and planning: consistent results in seconds, less manual tracing, and easier downstream work—so estimates, layouts, and conversations move forward with fewer delays.
More Projects
Computer Vision Systems
Built an AI-powered system for pre-construction floor plan drawings - capable of detecting different objects with high accuracy. We also developed a system to detect the speed of the cars in real time for automated highway monitoring.
Year :
2025
Industry :
Tech
Problem :
Architects, contractors, and DIY teams waste time tracing walls from photos or floor plans; results vary by person, and tricky inputs—dim lighting, skewed angles, or faded scans—cause misses that ripple into bad measurements, wrong material estimates, and rework later.
Solution :
I have created a Streamlit app that turns wall extraction into a one‑click flow: upload an image or plan, preview an instant overlay of detected walls, tweak sensitivity if needed, and export the result for takeoffs or CAD handoff; it standardizes detection quality and makes the task fast for non‑experts, too.
Challenge :
Delivering robust detection across messy real‑world inputs—from hand‑drawn sketches to smartphone photos—while keeping latency low in the browser, and gracefully handling edge cases like overlapping lines, reflective surfaces, and low‑dpi scans without overwhelming users with controls.
Summary :
A practical wall‑detection assistant for pre‑construction and planning: consistent results in seconds, less manual tracing, and easier downstream work—so estimates, layouts, and conversations move forward with fewer delays.
More Projects
Computer Vision Systems
Built an AI-powered system for pre-construction floor plan drawings - capable of detecting different objects with high accuracy. We also developed a system to detect the speed of the cars in real time for automated highway monitoring.
Year :
2025
Industry :
Tech
Problem :
Architects, contractors, and DIY teams waste time tracing walls from photos or floor plans; results vary by person, and tricky inputs—dim lighting, skewed angles, or faded scans—cause misses that ripple into bad measurements, wrong material estimates, and rework later.
Solution :
I have created a Streamlit app that turns wall extraction into a one‑click flow: upload an image or plan, preview an instant overlay of detected walls, tweak sensitivity if needed, and export the result for takeoffs or CAD handoff; it standardizes detection quality and makes the task fast for non‑experts, too.
Challenge :
Delivering robust detection across messy real‑world inputs—from hand‑drawn sketches to smartphone photos—while keeping latency low in the browser, and gracefully handling edge cases like overlapping lines, reflective surfaces, and low‑dpi scans without overwhelming users with controls.
Summary :
A practical wall‑detection assistant for pre‑construction and planning: consistent results in seconds, less manual tracing, and easier downstream work—so estimates, layouts, and conversations move forward with fewer delays.