HAL will be down for maintenance from Friday, June 10 at 4pm through Monday, June 13 at 9am. More information
Skip to Main content Skip to Navigation
Master Thesis

Le deep learning au service de la détection d’objets dans le cadre des relevés de façades

Abstract : Since 2012, we have seen the explosion in the use of deep learning in industries ranging from classification to image segmentation instance detection. Facade ortho-images are a type of geometrically rectified image used to draw facade elements, by using a CAD software. The Cascade-RCNN neural architecture is trained with multiple pre-cut mosaic ortho-images. The training consists in preparing ortho-images with detouring the elements making up a facade. Next, the trained RCNN model will make it possible to do predictions on new ortho-images. These predictions are composed of a clipping of the object in the form of a closed polygon without holes and a recognition confidence score. These will be converted to DXF in order to obtain layers and polylines of the facade elements. Thus we will allow the designer to save precious time.
Document type :
Master Thesis
Complete list of metadata

Contributor : Cnam - Service Commun de la Documentation - Esgt Connect in order to contact the contributor
Submitted on : Tuesday, June 22, 2021 - 2:28:28 PM
Last modification on : Monday, February 21, 2022 - 3:38:13 PM


 Restricted access
To satisfy the distribution rights of the publisher, the document is embargoed until : 2022-12-11

Please log in to resquest access to the document


  • HAL Id : dumas-03267362, version 1



Axel Marmajou. Le deep learning au service de la détection d’objets dans le cadre des relevés de façades. Sciences de l'ingénieur [physics]. 2020. ⟨dumas-03267362⟩



Record views