Unsupervised Semantic Segmentation Through Depth-Guided Feature Correlation and Sampling

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Sick, Leon, Engel, Dominik, Hermosilla, Pedro, Ropinski, Timo
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914728851275776
author Sick, Leon
Engel, Dominik
Hermosilla, Pedro
Ropinski, Timo
author_facet Sick, Leon
Engel, Dominik
Hermosilla, Pedro
Ropinski, Timo
contents Traditionally, training neural networks to perform semantic segmentation required expensive human-made annotations. But more recently, advances in the field of unsupervised learning have made significant progress on this issue and towards closing the gap to supervised algorithms. To achieve this, semantic knowledge is distilled by learning to correlate randomly sampled features from images across an entire dataset. In this work, we build upon these advances by incorporating information about the structure of the scene into the training process through the use of depth information. We achieve this by (1) learning depth-feature correlation by spatially correlate the feature maps with the depth maps to induce knowledge about the structure of the scene and (2) implementing farthest-point sampling to more effectively select relevant features by utilizing 3D sampling techniques on depth information of the scene. Finally, we demonstrate the effectiveness of our technical contributions through extensive experimentation and present significant improvements in performance across multiple benchmark datasets.
format Preprint
id arxiv_https___arxiv_org_abs_2309_12378
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Unsupervised Semantic Segmentation Through Depth-Guided Feature Correlation and Sampling
Sick, Leon
Engel, Dominik
Hermosilla, Pedro
Ropinski, Timo
Computer Vision and Pattern Recognition
Traditionally, training neural networks to perform semantic segmentation required expensive human-made annotations. But more recently, advances in the field of unsupervised learning have made significant progress on this issue and towards closing the gap to supervised algorithms. To achieve this, semantic knowledge is distilled by learning to correlate randomly sampled features from images across an entire dataset. In this work, we build upon these advances by incorporating information about the structure of the scene into the training process through the use of depth information. We achieve this by (1) learning depth-feature correlation by spatially correlate the feature maps with the depth maps to induce knowledge about the structure of the scene and (2) implementing farthest-point sampling to more effectively select relevant features by utilizing 3D sampling techniques on depth information of the scene. Finally, we demonstrate the effectiveness of our technical contributions through extensive experimentation and present significant improvements in performance across multiple benchmark datasets.
title Unsupervised Semantic Segmentation Through Depth-Guided Feature Correlation and Sampling
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2309.12378