Submitted by Shengyu Huang.

Submission data

Full nameMulti-task learner
DescriptionWe advocate multi-task learning as a way to improve scene recognition, building on the fact that the scene type is highly correlated with the objects in the scene, and therefore with its semantic segmentation into different object classes.
Publication titleIndoor Scene Recognition in 3D
Publication authorsShengyu Huang, Mikhail Usvyatsov, Konrad Schindler
Publication venueIROS 2020
Publication URLhttps://arxiv.org/abs/2002.12819
Input Data TypesUses Color,Uses Geometry        Uses 3D
Programming language(s)PyTorch
HardwareGeForce 1080Ti
Source code or download URLhttps://github.com/ShengyuH/Scene-Recognition-in-3D
Submission creation date18 Sep, 2019
Last edited25 Aug, 2020

Scene type classification results



Infoavg recallapartmentbathroombedroom / hotelbookstore / libraryconference roomcopy/mail roomhallwaykitchenlaundry roomliving room / loungemiscofficestorage / basement / garage
permissive0.7000.5001.0000.8820.5001.0001.0000.5001.0001.0000.7780.0000.9380.000