Cuepervision: self-supervised learning for continuous domain adaptation without catastrophic forgetting

2021 ◽  
Vol 106 ◽  
pp. 104079
Author(s):  
Mark Schutera ◽  
Frank M. Hafner ◽  
Jochen Abhau ◽  
Veit Hagenmeyer ◽  
Ralf Mikut ◽  
...  
Keyword(s):  
Supervised Learning ◽  
Domain Adaptation ◽  
Continuous Domain

scholarly journals U-shaped GAN for Semi-Supervised Learning and Unsupervised Domain Adaptation in High Resolution Chest Radiograph Segmentation

2022 ◽  
Vol 8 ◽  
Author(s):  
Hongyu Wang ◽  
Hong Gu ◽  
Pan Qin ◽  
Jia Wang
Keyword(s):  
Deep Learning ◽  
High Resolution ◽  
Supervised Learning ◽  
Chest Radiograph ◽  
Domain Adaptation ◽  
Medical Image Segmentation ◽  
Learning Approach ◽  
Feature Maps ◽  
Computational Burden ◽  
Unsupervised Domain Adaptation

Deep learning has achieved considerable success in medical image segmentation. However, applying deep learning in clinical environments often involves two problems: (1) scarcity of annotated data as data annotation is time-consuming and (2) varying attributes of different datasets due to domain shift. To address these problems, we propose an improved generative adversarial network (GAN) segmentation model, called U-shaped GAN, for limited-annotated chest radiograph datasets. The semi-supervised learning approach and unsupervised domain adaptation (UDA) approach are modeled into a unified framework for effective segmentation. We improve GAN by replacing the traditional discriminator with a U-shaped net, which predicts each pixel a label. The proposed U-shaped net is designed with high resolution radiographs (1,024 × 1,024) for effective segmentation while taking computational burden into account. The pointwise convolution is applied to U-shaped GAN for dimensionality reduction, which decreases the number of feature maps while retaining their salient features. Moreover, we design the U-shaped net with a pretrained ResNet-50 as an encoder to reduce the computational burden of training the encoder from scratch. A semi-supervised learning approach is proposed learning from limited annotated data while exploiting additional unannotated data with a pixel-level loss. U-shaped GAN is extended to UDA by taking the source and target domain data as the annotated data and the unannotated data in the semi-supervised learning approach, respectively. Compared to the previous models dealing with the aforementioned problems separately, U-shaped GAN is compatible with varying data distributions of multiple medical centers, with efficient training and optimizing performance. U-shaped GAN can be generalized to chest radiograph segmentation for clinical deployment. We evaluate U-shaped GAN with two chest radiograph datasets. U-shaped GAN is shown to significantly outperform the state-of-the-art models.

scholarly journals Cross Domain Adaptation of Crowd Counting with Model-Agnostic Meta-Learning

2021 ◽  
Vol 11 (24) ◽  
pp. 12037
Author(s):  
Xiaoyu Hou ◽  
Jihui Xu ◽  
Jinming Wu ◽  
Huaiyu Xu
Keyword(s):  
Feature Extraction ◽  
Supervised Learning ◽  
Density Estimation ◽  
Real World ◽  
Domain Adaptation ◽  
Manual Annotation ◽  
Crowd Counting ◽  
Cross Domain ◽  
Meta Learning ◽  
Density Map

Counting people in crowd scenarios is extensively conducted in drone inspections, video surveillance, and public safety applications. Today, crowd count algorithms with supervised learning have improved significantly, but with a reliance on a large amount of manual annotation. However, in real world scenarios, different photo angles, exposures, location heights, complex backgrounds, and limited annotation data lead to supervised learning methods not working satisfactorily, plus many of them suffer from overfitting problems. To address the above issues, we focus on training synthetic crowd data and investigate how to transfer information to real-world datasets while reducing the need for manual annotation. CNN-based crowd-counting algorithms usually consist of feature extraction, density estimation, and count regression. To improve the domain adaptation in feature extraction, we propose an adaptive domain-invariant feature extracting module. Meanwhile, after taking inspiration from recent innovative meta-learning, we present a dynamic-β MAML algorithm to generate a density map in unseen novel scenes and render the density estimation model more universal. Finally, we use a counting map refiner to optimize the coarse density map transformation into a fine density map and then regress the crowd number. Extensive experiments show that our proposed domain adaptation- and model-generalization methods can effectively suppress domain gaps and produce elaborate density maps in cross-domain crowd-counting scenarios. We demonstrate that the proposals in our paper outperform current state-of-the-art techniques.

scholarly journals StyHighNet: Semi-Supervised Learning Height Estimation from a Single Aerial Image via Unified Style Transferring

Sensors ◽  
2021 ◽  
Vol 21 (7) ◽  
pp. 2272
Author(s):  
Qian Gao ◽  
Xukun Shen
Keyword(s):  
Supervised Learning ◽  
Domain Adaptation ◽  
Superior Performance ◽  
Aerial Image ◽  
Distribution Maps ◽  
The Core ◽  
Public Data ◽  
Learning Modes ◽  
New City ◽  
The One

Recovering height information from a single aerial image is a key problem in the fields of computer vision and remote sensing. At present, supervised learning methods have achieved impressive results, but, due to domain bias, the trained model cannot be directly applied to a new scene. In this paper, we propose a novel semi-supervised framework, StyHighNet, for accurately estimating the height of a single aerial image in a new city that requires only a small number of labeled data. The core is to transfer multi-source images to a unified style, making the unlabeled data provide the appearance distribution as additional supervision signals. The framework mainly contains three sub-networks: (1) the style transferring sub-network maps multi-source images into unified style distribution maps (USDMs); (2) the height regression sub-network, with the function of predicting the height maps from USDMs; and (3) the style discrimination sub-network, used to distinguish the sources of USDMs. Among them, the style transferring sub-network shoulders dual responsibilities: On the one hand, it needs to compute USDMs with obvious characteristics, so that the height regression sub-network can accurately estimate the height maps. On the other hand, it is necessary that the USDMs have consistent distribution to confuse the style discrimination sub-network, so as to achieve the goal of domain adaptation. Unlike previous methods, our style distribution function is learned unsupervised, thus it is of greater flexibility and better accuracy. Furthermore, when the style discrimination sub-network is shielded, this framework can also be used for supervised learning. We performed qualitatively and quantitative evaluations on two sets of public data, Vaihingen and Potsdam. Experiments show that the framework achieved superior performance in both supervised and semi-supervised learning modes.

Sign in / Sign up

Export Citation Format

Share Document