Guest on 2D Is Against My Religion
Florent Poux
Point cloud researcher, founder of the 3D Geodata Academy, Neurones 3D
LinkedIn ↗Dr. Florent Poux is a point cloud researcher with a PhD on the Smart Point Cloud and a land surveying background, who runs the 3D Geodata Academy, consults on 3D AI projects and builds tools that turn scans into floor plans and IFC models. On 2DIAMR, Florent Poux showed live how a smart point cloud answers questions and exports an IFC model.
Episode
1 h 19 minPodcast
Nobody buys a point cloud: Florent Poux on AI that turns scans into IFC
Florent Poux · Neurones 3D
In Florent’s words
“The point cloud itself is not valuable for the end user, but it's valuable very much within this process.”
“This is a scientific principle of peer review. You never check the work yourself. You always have something else that check your work.”
“We tend to forget with hype that fundamentals are everything.”
Quoted in our tech guides
Will AI and robots replace 3D scanning operators and surveyors? What 11 practitioners say →
- “It's very dangerous to rely solely on AI.”▶ 16:39
How to sell 3D scanning and digital twins to clients: what 15 practitioners say →
- “With 2D deliverables, really the value lies in what you strip away.”▶ 18:09
Is a 3D scan a digital twin? What 10 practitioners say →
- “Having geometry, semantics or class information is not enough. You need to have relationships.”▶ 67:57
Scan-to-BIM: from point cloud to Revit model – what 8 practitioners say →
- “With the BIM part of things, it's in a matter of three seconds you have this entire IFC file.”▶ 39:00
Questions Florent answered
- Is manual point cloud segmentation still in demand?
- Yes – Florent Poux says there is massive demand, also visible among 3D Geodata Academy students, because segmentation lets you work at segment level instead of point level and speeds up every later step of a pipeline. ▶ 5:08
- How do you measure the accuracy of AI point cloud segmentation?
- Florent Poux advises checking precision (is a predicted chair really a chair) and recall (how many of the chairs are found), keeping both in balance and tying the metrics to the final deliverable, such as an IFC or OpenUSD model. ▶ 13:38
- Can AI generate an IFC model from a point cloud automatically?
- In the live demo, Florent Poux turns first-generation Matterport data into an IFC model in seconds; per Florent, the approach still struggles with cultural heritage (HBIM), where shapes differ from regular buildings. ▶ 36:29
- Should one AI model check another AI model’s work?
- Florent Poux compares it to scientific peer review – never check your own work – but warns that LLMs share similar architectures, so for QC you must be sure the checking AI is the right one. ▶ 43:44
- Does point cloud segmentation need RGB colours?
- Per Florent Poux, the Smart Point Cloud works on geometry alone if that is all there is, but uses intensity, RGB, 360° images and any other available data when present. ▶ 65:27
Talks about
Related guests
- Barry Bassnett – Digital twin practitioner, author of Tower of Twins
- Nikhil Sawlani – Founder, MultiSet AI
- Kelly Watt – Reality capture and digital twin programmes
- Arek Szadkowski – Global Lead, ArcGIS Reality, Esri
- Gregor Willhauck – Go-to-market, cloud for field systems (Trimble Connect), Trimble
- Michael Schoelkopf – Director, Product Management Reality Capture, NavVis