Submitted by ChinaUnicom -AI.

Submission data

Full name2D-3D Hybrid Convolutional Neural Network
DescriptionThis is a framework which combines 2D and 3D data to get accurate segmentation results.
Input Data TypesUses Color,Uses Geometry        Uses 2D,Uses 3D
Programming language(s)C++,PyTorch,Python
HardwareGeForce 2080Ti 8GB,i7-8700k
Submission creation date10 Dec, 2019
Last edited14 Jan, 2020

3D semantic label results

Infoavg ioubathtubbedbookshelfcabinetchaircountercurtaindeskdoorfloorotherfurniturepicturerefrigeratorshower curtainsinksofatabletoiletwallwindow
0.6930.5960.7890.8030.6770.8000.4690.8460.5540.5910.9480.5000.3160.6090.8470.7320.8080.5930.8940.8390.652