Learning to segment from misaligned and partial labels

Fobi, Simone; Conlon, Terence; Taneja, Jayant; Modi, Vijay

doi:10.1145/3378393.3402254

Cited by 6 publications

(2 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Existing work [17] uses OSM data to generate the annotations for buildings in satellite imagery. To correct the misaligned annotations from the OSM data, the method leverages a small set of manually corrected annotations to train an alignment correction network (ACN) before training a semantic segmentation model.…”

Section: Related Workmentioning

confidence: 99%

A Label Correction Algorithm Using Prior Information for Automatic and Accurate Geospatial Object Recognition

Duan¹,

Chiang²,

Leyk³

et al. 2021

Preprint

View full text Add to dashboard Cite

Thousands of scanned historical topographic maps contain valuable information covering long periods of time, such as how the hydrography of a region has changed over time. Efficiently unlocking the information in these maps requires training a geospatial objects recognition system, which needs a large amount of annotated data. Overlapping geo-referenced external vector data with topographic maps according to their coordinates can annotate the desired objects' locations in the maps automatically. However, directly overlapping the two datasets causes misaligned and false annotations because the publication years and coordinate projection systems of topographic maps are different from the external vector data. We propose a label correction algorithm, which leverages the color information of maps and the prior shape information of the external vector data to reduce misaligned and false annotations. The experiments show that the precision of annotations from the proposed algorithm is 10% higher than the annotations from a state-of-the-art algorithm. Consequently, recognition results using the proposed algorithm's annotations achieve 9% higher correctness than using the annotations from the state-of-the-art algorithm.

show abstract

Section: Related Workmentioning

confidence: 99%

A Label Correction Algorithm Using Prior Information for Automatic and Accurate Geospatial Object Recognition

Duan¹,

Chiang²,

Leyk³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…"Noise model based" approaches seek to estimate underlying noise structures in order to de-emphasize, relabel, or remove noisy labels so that the model does not learn from them, while "noise model free" approaches exploit noisy labels to improve robustness, for example, to speed up gradient descent through hard example mining or to avoid overfitting (Chang et al, 2017). There exist many different model architectures and loss functions for dealing with noisy labels across these categories (Mnih and Hinton, 2012;Fobi et al, 2020;Kang et al, 2020;Kang et al, 2021). This paper falls into the "noise model based" category, aiming to identify a noisy label distribution for potential relabeling.At the same time, as interpretability becomes a major focus in the field of Deep Learning, research on predictive uncertainty is on the rise, since critical applications of deep learning models require uncertainty measures such as confidence estimates to interpret and trust model predictions (Henne et al, 2020).…”

mentioning

confidence: 99%

Uncertainty is not sufficient for identifying noisy labels in training data for binary segmentation of building footprints

Ulman

Gütter²,

Niebling³

2023

Front. Remote Sens.

View full text Add to dashboard Cite

Obtaining high quality labels is a major challenge for the application of deep neural networks in the remote sensing domain. A common way of acquiring labels is the usage of crowd sourcing which can provide much needed training data sets but also often contains incorrect labels which can affect the training process of a deep neural network significantly. In this paper, we exploit uncertainty to identify a certain type of label noise for semantic segmentation of buildings in satellite imagery. That type of label noise is known as “omission noise,” i.e., missing labels for whole buildings which still appear in the satellite image. Following the literature, uncertainty during training can help in identifying the “sweet spot” between generalizing well and overfitting to label noise, which is further used to differentiate between noisy and clean labels. The differentiation between clean and noisy labels is based on pixel-wise uncertainty estimation and beta distribution fitting to the uncertainty estimates. For our study, we create a data set for building segmentation with different levels of omission noise to evaluate the impact of the noise level on the performance of the deep neural network during training. In doing so, we show that established uncertainty-based methods to identify noisy labels are in general not sufficient enough for our kind of remote sensing data. On the other hand, for some noise levels, we observe some promising differences between noisy and clean data which opens the possibility to refine the state-of-the-art methods further.

show abstract

Focused information learning method for change detection based on segmentation with limited annotations

Ahn,

Chung,

Park

et al. 2024

International Journal of Applied Earth Observation and Geoinfor

View full text Add to dashboard Cite

Learning to segment from misaligned and partial labels

Cited by 6 publications

References 11 publications

A Label Correction Algorithm Using Prior Information for Automatic and Accurate Geospatial Object Recognition

A Label Correction Algorithm Using Prior Information for Automatic and Accurate Geospatial Object Recognition

Uncertainty is not sufficient for identifying noisy labels in training data for binary segmentation of building footprints

Focused information learning method for change detection based on segmentation with limited annotations

Contact Info

Product

Resources

About