检索结果-内蒙古大学图书馆

ieee International Symposium on Biomedical Imaging

作者： Fengjun Zhao Yongfeng Chen Kaiming Huang Xiaowei He Xin Chen Yuqing Hou Xi’an Key Lab of Radiomics and Intelligent Perception School of Information Science and Technology Northwest University Xi’an China Department of Radiology Second Affiliated Hospital of Xi’an Jiaotong University Xi’an China

Segmentation allows localization and visualization of breast lesions of interest, thus essential for precise diagnosis, prognosis and treatment. However, accurate segmentation from magnetic resonance imaging (MRI) faces following challenges: 1) multiscale breast tumors with a large size range; 2) image quality degradation caused by motion artifacts; 3) blurred tumor boundaries especially malignant tumors; 4) limited labeled MR images. To address these challenges, we propose a semi-supervised boundary-guided segmentation network (Semi-BGSegNet) for breast tumors. Specifically, we first construct a boundary-guided baseline segmentation network (BGSegNet) that integrated multi-scale and attention modules to the encoder-decoder architecture. A dynamic boundary loss is combined with the cross-entropy loss to guide the update of BGSegNet. Then, we develop a lightweight pixel-level discriminator to discriminate predictions from labels. The discriminator can also provide pseudo labels for unlabeled data for semi-supervised training of BGSegNet. Experiments demonstrated the effectiveness of our model compared to other supervised and semi-supervised methods.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Mask Wearing Detection System Based On YOLO And Raspberry Pi

Mask Wearing Detection System Based On YOLO And Raspberry Pi

引用

Consumer Electronics and Computer Engineering (ICCECE), 2021 ieee International Conference on

作者： Ruishi Liang Yizhu Chen Shuaibing Li Huizhi Yang School of Computer University of Electronic Science and Technology of China Zhongshan Institute Zhongshan China

To identify mask wearing quickly and automatically in public places is particularly important for epidemic prevention and control. In this paper, we present a real-time mask wearing detection algorithm based on improved YOLOv5s, which speeds up the reasoning speed by 5~10% and achieves a detection accuracy of more than 96%. The proposed algorithm can be easily deployed in Raspberry PI. We also design a Web-based mask wearing detection system consisting of two parts: the cloud subsystem and the edge subsystem. The cloud part mainly realizes data storage, model training, equipment monitoring, big data visualization and other functions. The edge part uses Raspberry Pie as the core deployment equipment to complete data collection, model reasoning, information early warning and other functions. Our system features the advantages of high real-time, low cost and low network traffic. It can be widely deployed in the supermarket, parks, intelligent light poles and other open scenes, resulting in greater practical application value.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Av-Sepformer: Cross-Attention Sepformer for Audio-Visual Target Speaker Extraction

Av-Sepformer: Cross-Attention Sepformer for Audio-Visual Tar...

引用

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Jiuxin Lin Xinyu Cai Heinrich Dinkel Jun Chen Zhiyong Yan Yongqing Wang Junbo Zhang Zhiyong Wu Yujun Wang Helen Meng Shenzhen International Graduate School Tsinghua University Shenzhen China Xiaomi Inc Beijing China The Chinese University of Hong Kong Hong Kong SAR China

Visual information can serve as an effective cue for target speaker extraction (TSE) and is vital to improving extraction performance. In this paper, we propose AV-SepFormer, a SepFormer-based attention dual-scale model that utilizes cross- and self-attention to fuse and model features from audio and visual. AV-SepFormer splits the audio feature into a number of chunks, equivalent to the length of the visual feature. Then self- and cross-attention are employed to model the multi-modal features. Furthermore, we use a novel 2D positional encoding, that introduces the positional information between and within chunks and provides significant gains over the traditional positional encoding. Our model has two key advantages: the time granularity of audio chunked feature is synchronized to the visual feature, which alleviates the harm caused by the inconsistency of audio and video sampling rate; by combining self- and cross-attention, feature fusion and speech extraction processes are unified within an attention paradigm. The experimental results show that AV-SepFormer significantly outperforms other existing methods.

关键词： visualization Fuses Signal processing Feature extraction Transformers Encoding data mining

来源：评论

学校读者我要写书评

暂无评论

Super-Resolution for Macro X-Ray Fluorescence data Collected from Old Master Paintings

Super-Resolution for Macro X-Ray Fluorescence Data Collected...

引用

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Su Yarn Herman Verinaz-Jadan Jun-Jie Huang Nathan Daly Catherine Higgitt Pier Luigi Dragotti Department of Electrical and Electronic Engineering Imperial College London Department of Computer Science National University of Defense Technology China Scientific Department The National Gallery London

Macro X-ray fluorescence (MA-XRF) scanning is commonly used to non-invasively analyse Old Master paintings by mapping the distribution of the chemical elements present in the artworks. The visual quality of the element distribution maps is very important for characterising the materials and understanding the execution and condition of the painting. However, this quality is limited by the acquisition time for the XRF datacube, resulting in a trade-off between signal-to-noise ratio (SNR) and spatial resolution. To solve this problem we propose to enhance the spatial resolution of the XRF datacube of a painting leveraging a corresponding high-resolution (HR) RGB image. We achieve that by introducing a method based on coupled dictionary learning along with a similarity constraint based on mutual information. In particular, we divide the RGB image and the XRF datacube into a common part and a unique part based on whether the information is shared or not, and then transfer the HR information between the two common parts, resulting in high-quality reconstructions. Numerical results show that our XRF super-resolution method outperforms the other state-of-the-art approaches.

关键词： visualization Superresolution Machine learning Fluorescence Chemical elements Spatial resolution X-ray imaging

来源：评论

学校读者我要写书评

暂无评论

Research on the Interaction Design of Online Learning Platforms for University Students Based on a User’s Mental Model

Research on the Interaction Design of Online Learning Platfo...

引用

International Conference on Educational and Information Technology (ICEIT)

作者： Yujie Jiang Quandang Ma Lizhou Zhang Shihui Ma School of Undergraduate Wuhan University of Technology Wuhan China School of Navigation Wuhan University of Technology Wuhan China

Based on a user's mental model, the interaction (HCI) design strategy of online learning platforms for university students is explored to provide a theoretical basis for designing and developing online learning platforms. Through the theoretical overview, this study analyses the significance of the involvement of mental models in the interaction design of online learning platforms. A user's mental model for the interaction design of online learning platforms for university students is integrated. Through the quantitative and qualitative analysis of the questionnaire, the current problems and needs of university students in the interactive experience of online learning platforms are summarized. Combining a user's mental model and the questionnaire data, this study proposes an interaction design strategy for university students' online learning platform based on the mental model from three dimensions: the visual, the behavioral, and the reflective dimensions. It is hoped that it will meet the learning needs of university students in the online learning platform， and optimize their online learning experience.

关键词： Human computer interaction visualization Electronic learning Education Product design Cognitive science Behavioral sciences

来源：评论

学校读者我要写书评

暂无评论

Signal Preprocessing Towards IoT Acoustic data for Farm Pest Detection

Signal Preprocessing Towards IoT Acoustic Data for Farm Pest...

引用

Eurocon

作者： Destiny Kwabla Amenyedzi Micheline Kazeneza Anthony Vodacek Theofrida Julius Maginga Frederic Nzanywayingoma Philibert Nsengiyumva Peace Bamurigire Emmanuel Ndashimye African Centre of Excellence in Internet of Things University of Rwanda Kigali Rwanda Chester F. Carlson Center for Imaging Science Rochester Institute of Technology Rochester New York USA School of ICT University of Rwanda Kigali Rwanda Electrical and Electronics Engineering University of Rwanda Kigali Rwanda School of ECE Carnegie Mellon University Africa Kigali Rwanda

In Africa the realization of the Sustainable Development Goal 2 (SDG-2) of zero hunger by 2030 is threatened by agricultural pests. Acoustic technology offers one method for pest detection in agriculture. AudioMoth microphones were deployed in three farms in Rwanda. Band filters were applied to the files for listening to pest calls for labeling. The WAV files were processed using Scikit-maad package in Python to spectrograms, which allowed for visualization. The regions of interest were selected to be used for labeling. Results suggest that birds are the dominant pest during the morning sections of the day while frogs are active at night. The contribution of this work is to provide a roadmap towards labeling for detection of agricultural pests using machine learning at the edge.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Descriptive Analytics Analysis as Support for Ancestral Domain Preservation: Empirical Evidence Generation as support for Decision Making

Descriptive Analytics Analysis as Support for Ancestral Doma...

引用

ieee International Conference on Automatic Control and Intelligent Systems (I2CACIS)

作者： Eltimar T. Castro Nephi P. Romano Rev. Fr. Randy Jasper C. Odchigue Rev. Fr. John Christian U. Young Jannina de Jesus Alfante Rochelle A. Valencia Data Science Analytics Center Father Saturnino Urios University Butuan City Philippines Office of VP Academic Affairs and Research Father Saturnino Urios University Butuan City Philippines Office of the President dept. Father Saturnino Urios University Butuan City Philippines Arts and Science Program Father Saturnino Urios University Butuan City Philippines

Ancestral domain refers to the lands, territories, and resources possessed and administered collectively by indigenous peoples, such as the indigenous communities of the Philippines. Indigenous peoples who have inhabited and cared for these regions for generations attach great cultural, spiritual, and economic significance to them. Indigenous Peoples Rights Act (IPRA) of 1997 acknowledged for the first time the idea of ancestral domain in the Philippines. However, the IPRA’s implementation and the preservation of ancestral properties in the Philippines have encountered obstacles. There have been instances of unlawful logging, mining, and land appropriation in ancestral domains, typically with the participation of local government officials and influential persons. This has resulted in the uprooting of indigenous populations and the devastation of their land and resources. This research aimed to analyze the data in relation to tree planting activities initiated by several advocates, especially the Father Saturnino Urios University Foundation. Which aims, helping to preserve and saving Ancestral Land. This research employs data mining activity and develops insights according to generated patterns. Descriptive analytics is the branch of data analysis that involves summarizing and describing data. It typically involves using statistics, data visualization, and other techniques to identify patterns and trends in data. Descriptive analytics can be used to understand the characteristics of data and to communicate those characteristics to others. Further, the results show that tree planting activities conducted by some advocates resulted in good results, since planted trees near the watershed have had a survival rate of 92.28 (%) percent and around 43,270 seedlings planted in 88.05 hectares. On the other hand, abaca have been planted in the area for the livelihood of IP situated in the area, have a survival rate of 96 (%) percent. This result is the output of an ongoing reh

关键词：

来源：评论

学校读者我要写书评

暂无评论

Hearing and Seeing Abnormality: Self-Supervised Audio-Visual Mutual Learning for Deepfake Detection

Hearing and Seeing Abnormality: Self-Supervised Audio-Visual...

引用

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Chang-Sung Sung Jun-Cheng Chen Chu-Song Chen National Taiwan University Research Center for Information Technology Innovation Academia Sinica

The recent development of deepfakes has resulted in serious threats to society, such as spreading misinformation, defamation, etc. Although recent deepfake detection methods are capable of achieving satisfactory results for seen forgeries, the performance drops significantly for unseen ones. With proper supervised pretraining on auxiliary tasks as prior, the situation can be improved, but the requirement to collect a large number of additional annotations for these tasks may restrict the further development of a generalized deep-fake detector. To address this issue, we propose an Audio-Visual Temporal Synchronization for Deepfake Detection framework for detecting deepfakes that maintains reasonable detection capabilities for unseen ones. The primary objective of our framework is to determine whether there has been a forgery by evaluating the consistency between the sound and the faces in a video clip, together with the relationship between the two features. First, the spatiotemporal feature extraction network is pretrained in a self-supervised manner by exploiting the audio-visual temporal synchronization task to build up a rich representation based on the temporal synchronization relationship between the audio and its corresponding video. For pretraining, we use only real data and carefully selected negative samples with contrastive loss to train the model. A temporal classifier network is used to determine whether or not the video has been manipulated using the representations obtained from the pretrained feature extraction networks. To prevent the model from overfitting to certain manipulation-specific artifacts, we froze the feature extraction networks and only trained the final classifier network on forged data. Extensive experiments on unseen forgery categories and unseen datasets have shown the effectiveness of our method to achieve state-of-the-art results.

关键词： Training Deepfakes visualization Feature extraction Forgery data models Spatiotemporal phenomena

来源：评论

学校读者我要写书评

暂无评论

An Image Grid Can Be Worth a Video: Zero-Shot Video Question Answering Using a VLM

引用

ieee ACCESS 2024年 12卷 193057-193075页

作者： Kim, Wonkyun Choi, Changin Lee, Wonseok Rhee, Wonjong Seoul Natl Univ Dept Intelligence & Informat Seoul 08826 South Korea Seoul Natl Univ Interdisciplinary Program Artificial Intelligence Seoul 08826 South Korea Seoul Natl Univ AI Inst Seoul 08826 South Korea

Stimulated by the sophisticated reasoning capabilities of recent Large Language Models (LLMs), a variety of strategies for bridging video modality have been devised. A prominent strategy involves Video Language Models (VideoLMs), which train a learnable interface with video data to connect advanced vision encoders with LLMs. Recently, an alternative strategy has surfaced, employing readily available foundation models, such as VideoLMs and LLMs, across multiple stages for modality bridging. In this study, we introduce a simple yet novel strategy where only a single Vision Language Model (VLM) is utilized. Our starting point is the plain insight that a video comprises a series of images, or frames, interwoven with temporal information. The essence of video comprehension lies in adeptly managing the temporal aspects along with the spatial details of each frame. Initially, we transform a video into a single composite image by arranging multiple frames in our proposed grid layout. The resulting single image is termed as an image grid. This format, while maintaining the appearance of a solitary image, effectively retains temporal information within the grid structure at the pixel level. Therefore, the image grid approach enables direct application of a single high-performance VLM without necessitating any video-data training. Our extensive experimental analysis across ten zero-shot VQA benchmarks, including five open-ended and five multiple-choice benchmarks, reveals that the proposed Image Grid Vision Language Model (IG-VLM) surpasses the existing methods in nine out of ten benchmarks. We also discuss how IG-VLM can be extended for long videos and provide an extension method that consistently and reliably improves the performance. Our code is are available at: https://***/imagegridworth/IG-VLM

关键词： Benchmark testing Training Question answering (information retrieval) Cognition visualization data models Focusing Computational modeling Tuning Streaming media Image grid video question answering video representation vision language model

来源：评论

学校读者我要写书评

暂无评论

An Automatic Grading System for Neonatal Endotracheal Intubation with Multi-Task Convolutional Neural Network

An Automatic Grading System for Neonatal Endotracheal Intuba...

引用

ieee EMBS International Conference on Information Technology Applications in Biomedicine (ITAB)

作者： Yan Meng James K. Hahn Department of Computer Science The George Washington University Washington DC USA

Neonatal endotracheal intubation (ETI) is an intricate medical procedure that poses considerable challenges, demanding comprehensive training to effectively address potential complications in clinical practice. However, due to limited access to clinical opportunities, ETI training relies heavily on physical manikins to develop a certain level of competence before clinical exposure. Nonetheless, traditional training methods prove ineffective due to scarcity of expert instructors and the absence of internal situational awareness within the manikins, preventing thorough performance assessment for both trainees and instructors. To address this gap, there is a need to develop an automatic grading system that can assist trainees in performance assessment. In this paper, we proposed a multi-task Convolutional Neural Network (MTCNN) based model for assessing ETI proficiency, specifically targeting key performance features recommended by expert instructors. The model comprises three modules: an ETI simulation module that captures the ETI procedures performed on a standard neonatal task trainer manikin, an automatic grading module that extracts and grades the identified key performance features, and a data visualization module that presents the assessment results in a user-friendly manner. The experimental results demonstrated that the proposed automatic grading system achieved an average classification accuracy of 93.6%. This study established the successful integration of intuitive observed features with latent features derived from multivariate time series (MTS) data, coupled with multi-task deep learning techniques, for the automatic assessment of ETI performance.

关键词：

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：