PiCoCo: Pixelwise Contrast and Consistency Learning for Semisupervised Building Footprint Segmentation

Kang, Jian; Wang, Zhirui; Zhu, Ruoxin; Sun, Xian; Fernández-Beltrán, Rubén; Plaza, Antonio

doi:10.1109/jstars.2021.3119286

Cited by 45 publications

(35 citation statements)

References 51 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To the best of our knowledge, the proposed CPR strategy is the first example of using semi-supervised learning for forest variable prediction, tree height in our study case. To estimate its potential, two well-known semi-supervised strategies previously used for segmentation/change-detection [27], [45] are converted to regression task as additional baseline models. They include reconstruction-based two-step training and Siamese-Network-based consistency learning.…”

Section: B Baseline Modelsmentioning

confidence: 99%

Improved Semisupervised UNet Deep Learning Model for Forest Height Mapping With Satellite SAR and Optical Data

et al. 2022

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

In this study, we introduce an improved semisupervised deep learning approach, and demonstrate its suitability for modeling the relationship between forest structural parameters and satellite remote sensing imagery and producing forest maps. The improved approach is based on a popular UNet model, modified and fine-tuned to improve the forest parameter prediction performance. Within the improved model, squeezeand-excitation blocks are embedded to re-calibrate the multisource features via retrieved channel-wise self-attention and a novel Cross-Pseudo Regression strategy is implemented to train the model in a semi-supervised way. The improvement imposes consistency learning on two perturbed network branches, generating regression pseudo-reference and expanding the dataset size. For demonstration, we used satellite synthetic aperture radar (SAR) Sentinel-1 and multispectral optical Sentinel-2 images as remote sensing data, complemented with reference data represented by forest tree height as one of the key forest structural variables. The study area is located in a boreal forestland in Central Finland. Proposed approach showed larger accuracy compared to traditional machine learning methods such as random forests and boosting trees, and baseline UNet model. Best accuracy figures for forest tree height were achieved with combined SAR and optical imagery and were as small as 24.1 % RMSE (root mean square error) on pixel-level and 15.4 % RMSE on forest stand level.

show abstract

Section: B Baseline Modelsmentioning

confidence: 99%

Improved Semisupervised UNet Deep Learning Model for Forest Height Mapping With Satellite SAR and Optical Data

et al. 2022

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

show abstract

“…Pixelwise Contrast and Consistency Learning (PiCoCo), proposed by Kang et al [25], seeks consistency in closed set semantic segmentation using a joint loss function that is summation of a supervised loss term, a contrast loss term, and a consistency loss term. The supervised is a standard semantic segmentation loss term composed by a Cross Entropy and a Dice loss term; for the contrastive loss a selection of positive and negative samples enforce the model to improve its generalization capabilities; the consistency loss term consists in a summation of a cross entropy and a dice loss of heavy augmented pairs of input and labels to enforce semantic consistency and robustness to the learning process.…”

Section: Related Work a Semantic Consistency In Segmentationmentioning

confidence: 99%

FuSS: Fusing Superpixels for Improved Segmentation Consistency

Nunes¹,

Pereira²,

Oliveira³

et al. 2022

Preprint

View full text Add to dashboard Cite

In this work, we propose two different approaches to improve the semantic consistency of Open Set Semantic Segmentation. First, we propose a method called OpenGMM that extends the OpenPCS framework using a Gaussian Mixture of Models to model the distribution of pixels for each class in a multimodal manner. The second approach is a post-processing which uses superpixels to enforce highly homogeneous regions to behave equally, rectifying erroneous classified pixels within these regions, we also proposed a novel superpixel method called FuSS. All tests were performed on ISPRS Vaihingen and Potsdam datasets, and both methods were capable to improve quantitative and qualitative results for both datasets. Besides that, the post-process with FuSS achieved state-of-the-art results for both datasets. The official implementation is available at: https://github.com/iannunes/FuSS.

show abstract

“…Besides, water is also an integral part of different thematic and topographic maps used for many different purposes. Under this scenario, timely updated data are logically required to effectively monitor water bodies, which tend to change from time to time unlike other more stable structures like buildings or roads [6]. Unfortunately, this demand is difficult to cover using time consuming in-situ procedures, specially in the context of developing countries [7].…”

Section: Introductionmentioning

confidence: 99%

Attentional Dense Convolutional Neural Network for Water Body Extraction From Sentinel-2 Images

Parajuli

Fernández-Beltrán

Kang

et al. 2022

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

Self Cite

View full text Add to dashboard Cite

Monitoring water bodies from remote sensing data is certainly an essential task to supervise the actual conditions of the available water resources for environment conservation, sustainable development and many other applications. Being Sentinel-2 images some of the most attractive data, existing traditional index-based and deep learning-based water extraction methods still have important limitations to effectively deal with large heterogeneous areas since many types of water bodies with different spatial-spectral complexities are logically expected. Note that, in this scenario, optimal feature abstraction and neighbourhood information may certainly vary from water to water pixel, however existing methods are generally constrained by a fix abstraction level and amount of land cover context. To address these issues, this paper presents a new attentional dense convolutional neural network (AD-CNN) specially designed for water body extraction from Sentinel-2 imagery. On the one hand, AD-CNN exploits dense connections to allow uncovering deeper features while simultaneously characterizing multiple data complexities. On the other hand, the proposed model also implements a new residual attention module to dynamically put the focus on the most relevant spatial-spectral features for classifying water pixels. To test the performance of AD-CNN, a new water database of Nepal (WaterPAL) is also built. The conducted experiments reveal the competitive performance of the proposed architecture with respect to several traditional indexbased and state-of-the-art deep learning-based water extraction models. The codes and data related to this paper will be accessible on https://github.com/rufernan/ADCNN.

show abstract

PiCoCo: Pixelwise Contrast and Consistency Learning for Semisupervised Building Footprint Segmentation

Cited by 45 publications

References 51 publications

Improved Semisupervised UNet Deep Learning Model for Forest Height Mapping With Satellite SAR and Optical Data

Improved Semisupervised UNet Deep Learning Model for Forest Height Mapping With Satellite SAR and Optical Data

FuSS: Fusing Superpixels for Improved Segmentation Consistency

Attentional Dense Convolutional Neural Network for Water Body Extraction From Sentinel-2 Images

Contact Info

Product

Resources

About