This paper presents a system towards the generation of multi-label datasets
from web data in an unsupervised manner. To achieve this objective, this work
comprises two main contributions, namely: a) the generation of a low-noise
unsupervised single-label dataset from web-data, and b) the augmentation of
labels in such dataset (from single label to multi label). The generation of a
single-label dataset uses an unsupervised noise reduction phase (clustering and
selection of clusters using anchors) obtaining a 85% of correctly labeled
images. An unsupervised label augmentation process is then performed to assign
new labels to the images in the dataset using the class activation maps and the
uncertainty associated with each class. This process is applied to the dataset
generated in this paper and a public dataset (Places365) achieving a 9.5% and
27% of extra labels in each dataset respectively, therefore demonstrating that
the presented system can robustly enrich the initial dataset.
Description
[2005.05623] Unsupervised Multi-label Dataset Generation from Web Data
%0 Generic
%1 roig2020unsupervised
%A Roig, Carlos
%A Varas, David
%A Masuda, Issey
%A Riveiro, Juan Carlos
%A Bou-Balust, Elisenda
%D 2020
%K data generation set web
%T Unsupervised Multi-label Dataset Generation from Web Data
%U http://arxiv.org/abs/2005.05623
%X This paper presents a system towards the generation of multi-label datasets
from web data in an unsupervised manner. To achieve this objective, this work
comprises two main contributions, namely: a) the generation of a low-noise
unsupervised single-label dataset from web-data, and b) the augmentation of
labels in such dataset (from single label to multi label). The generation of a
single-label dataset uses an unsupervised noise reduction phase (clustering and
selection of clusters using anchors) obtaining a 85% of correctly labeled
images. An unsupervised label augmentation process is then performed to assign
new labels to the images in the dataset using the class activation maps and the
uncertainty associated with each class. This process is applied to the dataset
generated in this paper and a public dataset (Places365) achieving a 9.5% and
27% of extra labels in each dataset respectively, therefore demonstrating that
the presented system can robustly enrich the initial dataset.
@misc{roig2020unsupervised,
abstract = {This paper presents a system towards the generation of multi-label datasets
from web data in an unsupervised manner. To achieve this objective, this work
comprises two main contributions, namely: a) the generation of a low-noise
unsupervised single-label dataset from web-data, and b) the augmentation of
labels in such dataset (from single label to multi label). The generation of a
single-label dataset uses an unsupervised noise reduction phase (clustering and
selection of clusters using anchors) obtaining a 85% of correctly labeled
images. An unsupervised label augmentation process is then performed to assign
new labels to the images in the dataset using the class activation maps and the
uncertainty associated with each class. This process is applied to the dataset
generated in this paper and a public dataset (Places365) achieving a 9.5% and
27% of extra labels in each dataset respectively, therefore demonstrating that
the presented system can robustly enrich the initial dataset.},
added-at = {2020-09-28T10:32:25.000+0200},
author = {Roig, Carlos and Varas, David and Masuda, Issey and Riveiro, Juan Carlos and Bou-Balust, Elisenda},
biburl = {https://www.bibsonomy.org/bibtex/25a423be623ec5649610b1989cc6457f4/parismic},
description = {[2005.05623] Unsupervised Multi-label Dataset Generation from Web Data},
interhash = {6c4aa306722f567af1aa8b29c404d60e},
intrahash = {5a423be623ec5649610b1989cc6457f4},
keywords = {data generation set web},
note = {cite arxiv:2005.05623Comment: The 3rd Workshop on Visual Understanding by Learning from Web Data 2019},
timestamp = {2020-09-28T10:32:25.000+0200},
title = {Unsupervised Multi-label Dataset Generation from Web Data},
url = {http://arxiv.org/abs/2005.05623},
year = 2020
}