In this paper, we introduce a realistic and challenging domain adaptation problem called Universal Semi-supervised Model Adaptation (USMA), which i) requires only a pre-trained source model, ii) allows the source and ...
详细信息
ISBN:
(纸本)9798350318920;9798350318937
In this paper, we introduce a realistic and challenging domain adaptation problem called Universal Semi-supervised Model Adaptation (USMA), which i) requires only a pre-trained source model, ii) allows the source and target domain to have different label sets, i.e., they share a common label set and hold their own private label set, and iii) requires only a few labeled samples in each class of the target domain. To address USMA, we propose a collaborative consistency training framework that regularizes the prediction consistency between two models, i.e., a pretrained source model and its variant pre-trained with target data only, and combines their complementary strengths to learn a more powerful model. The rationale of our framework stems from the observation that the source model performs better on common categories than the target-only model, while on target-private categories, the target-only model performs better. We also propose a two-perspective, i.e., sample-wise and class-wise, consistency regularization to improve the training. Experimental results demonstrate the effectiveness of our method on several benchmark datasets.
A widely studied problem in computer science is the restoration, segmentation, and classification of images, which involves imageprocessing, computer vision, and machine learning techniques. Deep learning has made si...
详细信息
The proceedings contain 127 papers. The topics discussed include: Advanced data storage and processing technologies in a next-generation electric information acquisition system;analyzing file access characteristics fo...
ISBN:
(纸本)9798350355253
The proceedings contain 127 papers. The topics discussed include: Advanced data storage and processing technologies in a next-generation electric information acquisition system;analyzing file access characteristics for deep learning workloads on mobile devices;optimal scheduling of distributed energy storage for electric vehicles based on evolutionary dissipation theory;a novel semi-supervised learning approach for referring expression comprehension;research and implementation of material image subject segmentation method based on machinevision;application of image recognition and 3D reconstruction technology in virtual museum system;knowledge graph technology-based active research and judgment technology for electric power customer complaint risk;and path planning for unmanned underwater vehicles based on improved ant colony algorithm.
A generic fundus foreground extractor is required for the standardization of fundus datasets in machine-learning applications due to the vast range of retinal fundus images. Some fundus images have a large amount of n...
详细信息
Simulator sickness induced by 360 & DEG;stereoscopic video contents is a prolonged challenging issue in Virtual Reality (VR) system. Current machine learning models for simulator sickness prediction ignore the und...
详细信息
Simulator sickness induced by 360 & DEG;stereoscopic video contents is a prolonged challenging issue in Virtual Reality (VR) system. Current machine learning models for simulator sickness prediction ignore the underlying interdependencies and correlations across multiple visual features which may lead to simulator sickness. We propose a model for sickness prediction by automatic learning and adaptive integrating multi-level mappings from stereoscopic video features to simulator sickness scores. Firstly, saliency, optical flow and disparity features are extracted from videos to reflect the factors causing simulator sickness, including human attention area, motion velocity and depth information. Then, these features are embedded and fed into a 3-dimensional convolutional neural network (3D CNN) to extract the underlying multi-level knowledge which includes low-level and higher-order visual concepts, and global image descriptor. Finally, an attentional mechanism is exploited to adaptively fuse multi-level information with attentional weights for sickness score estimation. The proposed model is trained by an end-to-end approach and validated over a public dataset. Comparison results with state-of-the-art models and ablation studies demonstrated improved performance in terms of Root Mean Square Error (RMSE) and Pearson Linear Correlation Coefficient.
image restoration, a critical task in computer vision and imageprocessing, focuses on recovering degraded or damaged images to their original, high-quality state. This paper introduces an innovative approach to image...
详细信息
Computer vision is one of the important areas and directions of deep learning research, which requires different approaches to be chosen for different fields due to the complexity and diversity of vision tasks. In the...
详细信息
ISBN:
(纸本)9781665464680
Computer vision is one of the important areas and directions of deep learning research, which requires different approaches to be chosen for different fields due to the complexity and diversity of vision tasks. In the field of aviation, the existing image resources are still far from the real needs due to the influence and constraints of realistic scenes and difficulties of image acquisition. More detailed and comprehensive images can better provide reliable technical support and basis for applications, and then make more accurate decisions on problems, which requires generating more effective images to expand the data. Generative Adversarial Networks (GAN) are the fastest growing and most effective generation method in recent years, so this experiment investigates the application of GAN on aviation data, taking images of airplanes, cars and ships as examples to conduct quantitative research. The effect on the effect of GAN is studied from the perspective of image size, number of images, number of iterations, and different categories of images, in order to obtain better parameter settings for generating effective images, which provides a theoretical and experimental basis for the subsequent application of GAN in the aviation field to generate more images with similar characteristics and solve the problem of insufficient data.
As a very important branch of computer science and engineering, graphics, and imageprocessing is a research topic of capturing, storing, and manipulating information from reflected electromagnetic waves from objects ...
详细信息
This study addresses the pressing need for computer systems to interpret digital media images with a level of sophistication comparable to human visual perception. By leveraging Convolutional Neural Networks (CNNs), w...
详细信息
1. Animal phenotypic traits are utilised in a variety of studies. Often the traits are measured from images. The processing of a large number of images can be challenging;nevertheless, image analytical applications, b...
详细信息
1. Animal phenotypic traits are utilised in a variety of studies. Often the traits are measured from images. The processing of a large number of images can be challenging;nevertheless, image analytical applications, based on neural networks, can be an effective tool in automatic trait collection.2. Our aim was to develop a stand-alone application to effectively segment an arthropod from an image and to recognise individual body parts: namely, head, thorax (or prosoma), abdomen and four pairs of appendages. It is based on convolutional neural network with U-Net architecture trained on more than a thousand images showing dorsal views of arthropods (mainly of wingless insects and spiders). The segmentation model gave very good results, with the automatically generated segmentation masks usually requiring only slight manual adjustments.3. The application, named MAPHIS, can further (1) organise and preprocess the images;(2) adjust segmentation masks using a simple graphical editor;and (3) calculate various size, shape, colouration and pattern measures for each body part organised in a hierarchical manner. In addition, a special plug-in function can align body profiles of selected individuals to match a median profile and enable comparison among groups. The usability of the application is shown in three practical examples.4. The application can be used in a variety of fields where measures of phenotypic diversity are required, such as taxonomy, ecology and evolution (e.g. mimetic similarity). Currently, the application is limited to arthropods, but it can be easily extended to other animal taxa.
暂无评论