Member-only story

How to Build an Evacuation Map From a 3D Scan (Python)

The complete guide to measuring walking distances, escape routes and blocked-exit scenarios from a 3D point cloud scan, step by step in Python

25 min readOct 26, 2026

A before and after thumbnail on a light sepia card with a deep blue border, a top view of an office floor point cloud on the left and a grey floor plan shaded by walking distance with red escape routes running to the exits on the right, under the headline 3D evacuation plan. — The scanned office floor before, the walking distance to the nearest exit after. © F. Poux

Fires. Power line hazards. Earthquakes.

With climate change, emergencies are not rare events we plan for once and forget. They are scenarios we have to handle better, and faster, every time.

I am Florent Poux, Ph.D. in 3D point clouds and founder of the 3D Geodata Academy, and I have spent 15 years building 3D and AI systems that help people act on what a scan shows.

This one is very dear to my heart.

It builds on everything I have developed over those 15 years, and it is the work I do today with companies I love, companies that create solutions for the moments things go wrong.

So let me share the knowledge and know-how behind the scenes:

I share the core process as open knowledge, so that you can build it too.

You do not need my background to follow it, and you do not need to be a coder: with AI agents building alongside you, the pace is like nothing we have seen before. (Want the fast track…

Really cool work. I wonder how the results would change if you assume different traffic density distributions over the space. For example, imagine a museum or concert venue where people tend to congregate in specific places where just using the…

this line should be on a poster. defaulting every unknown class to obstacle is the same idea as <fail safe defaults> in security design, and it's great to see it applied to people, not just code, Florent...

Data Science Collective

How to Build an Evacuation Map From a 3D Scan (Python)

The complete guide to measuring walking distances, escape routes and blocked-exit scenarios from a 3D point cloud scan, step by step in Python

Written by Florent Poux, Ph.D.⁠

Published in Data Science Collective

Responses (5)

More from Florent Poux, Ph.D. and Data Science Collective

How to Turn a 3D Scan Into a CAD Floor Plan

The step-by-step guide to turning a 3D point cloud scan into an editable DXF + PDF floor plan with clean vector walls

Is Running a Local LLM Still Worth It After Haiku 5.5?

A cheap cloud model now matches Qwen 3.8 27B. What that does to the 5 usual reasons for running a model on your own GPU.

What Is the Best Local LLM for Mac in 2026? (M1 to M6)

A 2021 M1 Pro writes answers faster than a new M5 Air. The best model for every Mac from 8GB to 512GB, plus voice, embeddings and image.

3D Scan-to-BIM 101: Hands-on Automation Course

A practical 3D Scan-to-BIM guide with a focus on automated reconstruction of floors, ceilings, and walls, then exporting it as an .IFC file

Recommended from Medium

What Is the Best Local LLM for Mac in 2026? (M1 to M6)

A 2021 M1 Pro writes answers faster than a new M5 Air. The best model for every Mac from 8GB to 512GB, plus voice, embeddings and image.

8 New AI Terms You Should Know to Be Ahead of 90% of People

A few more AI terms to know beyond tokens, prompts, and hallucinations.

An Introduction to Markov Chains

What they are, how they work, and why you’ll see them everywhere from Google search to LLMs to modern marketing analytics

The Best Budget Local Inference Machine Ships Next Month — Here’s Why It’s Worth the Wait

⚠️ Update (November 2026): This post is from May 2026. The hardware landscape has changed significantly — Neural Accelerators, corrected…

Autoencoders: Learning Useful Representations Without Labels

Machine learning often depends on labelled data.

Advanced RAG: Layout Analysis for Document Ingestion

Locating text, tables, images, and preserving the layout clues that connect them.

Responses (5)

More from Florent Poux, Ph.D. and Data Science Collective

How to Turn a 3D Scan Into a CAD Floor Plan

The step-by-step guide to turning a 3D point cloud scan into an editable DXF + PDF floor plan with clean vector walls

Is Running a Local LLM Still Worth It After Haiku 5.5?

A cheap cloud model now matches Qwen 3.8 27B. What that does to the 5 usual reasons for running a model on your own GPU.

What Is the Best Local LLM for Mac in 2026? (M1 to M6)

A 2021 M1 Pro writes answers faster than a new M5 Air. The best model for every Mac from 8GB to 512GB, plus voice, embeddings and image.

3D Scan-to-BIM 101: Hands-on Automation Course

A practical 3D Scan-to-BIM guide with a focus on automated reconstruction of floors, ceilings, and walls, then exporting it as an .IFC file

Recommended from Medium

What Is the Best Local LLM for Mac in 2026? (M1 to M6)

A 2021 M1 Pro writes answers faster than a new M5 Air. The best model for every Mac from 8GB to 512GB, plus voice, embeddings and image.

8 New AI Terms You Should Know to Be Ahead of 90% of People

A few more AI terms to know beyond tokens, prompts, and hallucinations.

An Introduction to Markov Chains

What they are, how they work, and why you’ll see them everywhere from Google search to LLMs to modern marketing analytics

The Best Budget Local Inference Machine Ships Next Month — Here’s Why It’s Worth the Wait

⚠️ Update (November 2026): This post is from May 2026. The hardware landscape has changed significantly — Neural Accelerators, corrected…

Autoencoders: Learning Useful Representations Without Labels

Machine learning often depends on labelled data.

Advanced RAG: Layout Analysis for Document Ingestion

Locating text, tables, images, and preserving the layout clues that connect them.