EdgeNet: Semantic Scene Completion from RGB-D images

08/08/2019
by   Aloisio Dourado, et al.
4

Semantic scene completion is the task of predicting a complete 3D representation of volumetric occupancy with corresponding semantic labels for a scene from a single point of view. Previous works on Semantic Scene Completion from RGB-D data used either only depth or depth with colour by projecting the 2D image into the 3D volume resulting in a sparse data representation. In this work, we present a new strategy to encode colour information in 3D space using edge detection and flipped truncated signed distance. We also present EdgeNet, a new end-to-end neural network architecture capable of handling features generated from the fusion of depth and edge information. Experimental results show improvement of 6.9 end-to-end approaches.

READ FULL TEXT

page 2

page 3

page 4

page 5

page 7

research
02/13/2018

Semantic Scene Completion Combining Colour and Depth: preliminary experiments

Semantic scene completion is the task of producing a complete 3D voxel r...
research
03/23/2020

Atlas: End-to-End 3D Scene Reconstruction from Posed Images

We present an end-to-end 3D reconstruction method for a scene by directl...
research
03/17/2023

Semantic Scene Completion with Cleaner Self

Semantic Scene Completion (SSC) transforms an image of single-view depth...
research
11/26/2021

Data Augmented 3D Semantic Scene Completion with 2D Segmentation Priors

Semantic scene completion (SSC) is a challenging Computer Vision task wi...
research
09/03/2019

Counterfactual Depth from a Single RGB Image

We describe a method that predicts, from a single RGB image, a depth map...
research
03/16/2023

Cross-Dimensional Refined Learning for Real-Time 3D Visual Perception from Monocular Video

We present a novel real-time capable learning method that jointly percei...
research
03/31/2020

Attention-based Multi-modal Fusion Network for Semantic Scene Completion

This paper presents an end-to-end 3D convolutional network named attenti...

Please sign up or login with your details

Forgot password? Click here to reset