Submitted by Lisa Weijler.

Submission data

Full nameTest-Time Training for Indoor 3D Semantic Segmentation
Publication titleTTT-KD: Test-Time Training for 3D Semantic Segmentation through Knowledge Distillation from Foundation Models
Publication authorsLisa Weijler, Muhammad Jehanzeb Mirza, Leon Sick, Can Ekkazan, Pedro Hermosilla
Publication URLhttps://arxiv.org/abs/2403.11691
Input Data TypesUses Color        Uses 2D,Uses 3D
Programming language(s)Python, Cuda
HardwareA6000
Submission creation date15 Nov, 2023
Last edited11 Apr, 2024
Last uploaded15 Nov, 2023

3D semantic label results

Infoavg ioubathtubbedbookshelfcabinetchaircountercurtaindeskdoorfloorotherfurniturepicturerefrigeratorshower curtainsinksofatabletoiletwallwindow
0.7730.6460.8180.8090.7740.8780.5810.9430.6870.7040.9780.6070.3360.7750.9120.8380.8230.6940.9670.8990.794