SLIDE 1
Javier VENEGAS javier.venegas@novalian.fr Paris, France
Jean-Paul BALOCHE CEO Constantin CHEVANCE Head of Product Sergejs MIHAILOVS Data Engineer Diego ZUNIGA Data Scientist
Javier VENEGAS CTO
Understanding Buildings using Machine Learning Javier VENEGAS - - PowerPoint PPT Presentation
Understanding Buildings using Machine Learning Javier VENEGAS - - PowerPoint PPT Presentation
Understanding Buildings using Machine Learning Javier VENEGAS javier.venegas@novalian.fr Paris, France OUR TEAM Jean-Paul BALOCHE Constantin CHEVANCE CEO Head of Product Javier VENEGAS Sergejs MIHAILOVS Diego ZUNIGA CTO Data Engineer
SLIDE 2
SLIDE 3
BIM : INDUSTRIALIZATION TIME
70’s
Draw complex buildings with strong constraints Draw complex buildings with strong constraints
Now
SLIDE 4
BIM : INDUSTRIALIZATION TIME Standard exchange data model : IFC PDF for BIM
SLIDE 5
EXAMPLE Facade is one only wall, related to one floor. Take others floors = no wall GENERAL GEOMETRIC DATA INCONSISTENCY
SLIDE 6
OBJECT GEOMETRIC DATA INCONSISTENCY
Thi his s is is not not a a doo
- or.
EXAMPLE
- Language
- Measuring unit
- Definition of a door
- Modelisation length vs Metadata length
- Depth of the door = depth of the wall ?
- ...
SLIDE 7
BIM Platform & APP DeepBIM Transfo 3D Graph DOC SQL OUR SOLUTION : DeepBIM
DATA WRANGLING
Neural Networks
SLIDE 8
API Gateway Project API Queue Handler Kubernetes Cluster APIs :
- 3D retranscription
- Graph Creation
- IPC Ressources
- ...
Deep Learning Neural Network DeepBIM : ARCHITECTURE
SLIDE 9
IFC Documents DeepBIM : BUILDING INFORMATION MODEL 3D Transformation
Exploration
SLIDE 10
DeepBIM : FROM IMPLICIT DATA TO EXPLICIT DATA
SLIDE 11
DeepBIM : CREATING A NEW 3D REPRESENTATION
SLIDE 12
DeepBIM : DETECTING SPACES
SLIDE 13
DeepBIM : THE NEW DATA REPRESENTATION
SLIDE 14
DATA EXPLORATION EXAMPLE
SLIDE 15
Dashboard example
SLIDE 16