Soft labelling for semantic segmentation: Bringing coherence to label down-sampling

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Alcover-Couso, Roberto, Escudero-Vinolo, Marcos, SanMiguel, Juan C., Martinez, Jose M.
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913236639547392
author Alcover-Couso, Roberto
Escudero-Vinolo, Marcos
SanMiguel, Juan C.
Martinez, Jose M.
author_facet Alcover-Couso, Roberto
Escudero-Vinolo, Marcos
SanMiguel, Juan C.
Martinez, Jose M.
contents In semantic segmentation, training data down-sampling is commonly performed due to limited resources, the need to adapt image size to the model input, or improve data augmentation. This down-sampling typically employs different strategies for the image data and the annotated labels. Such discrepancy leads to mismatches between the down-sampled color and label images. Hence, the training performance significantly decreases as the down-sampling factor increases. In this paper, we bring together the down-sampling strategies for the image data and the training labels. To that aim, we propose a novel framework for label down-sampling via soft-labeling that better conserves label information after down-sampling. Therefore, fully aligning soft-labels with image data to keep the distribution of the sampled pixels. This proposal also produces reliable annotations for under-represented semantic classes. Altogether, it allows training competitive models at lower resolutions. Experiments show that the proposal outperforms other down-sampling strategies. Moreover, state-of-the-art performance is achieved for reference benchmarks, but employing significantly less computational resources than foremost approaches. This proposal enables competitive research for semantic segmentation under resource constraints.
format Preprint
id arxiv_https___arxiv_org_abs_2302_13961
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Soft labelling for semantic segmentation: Bringing coherence to label down-sampling
Alcover-Couso, Roberto
Escudero-Vinolo, Marcos
SanMiguel, Juan C.
Martinez, Jose M.
Computer Vision and Pattern Recognition
In semantic segmentation, training data down-sampling is commonly performed due to limited resources, the need to adapt image size to the model input, or improve data augmentation. This down-sampling typically employs different strategies for the image data and the annotated labels. Such discrepancy leads to mismatches between the down-sampled color and label images. Hence, the training performance significantly decreases as the down-sampling factor increases. In this paper, we bring together the down-sampling strategies for the image data and the training labels. To that aim, we propose a novel framework for label down-sampling via soft-labeling that better conserves label information after down-sampling. Therefore, fully aligning soft-labels with image data to keep the distribution of the sampled pixels. This proposal also produces reliable annotations for under-represented semantic classes. Altogether, it allows training competitive models at lower resolutions. Experiments show that the proposal outperforms other down-sampling strategies. Moreover, state-of-the-art performance is achieved for reference benchmarks, but employing significantly less computational resources than foremost approaches. This proposal enables competitive research for semantic segmentation under resource constraints.
title Soft labelling for semantic segmentation: Bringing coherence to label down-sampling
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2302.13961