检索结果-内蒙古大学图书馆

Document Image Analysis Using Deep Multi-modular Features

SN Computer Science 2023年第1期4卷 1-19页

作者： Jobin, K.V. Mondal, Ajoy Jawahar, C.V. Center for Visual Information Technology International Institute Information Technology Telangana Hyderabad 500032 India

Texture or repeating patterns, discriminative patches, and shapes are the salient features for various document image analysis problems. This article proposes a deep network architecture that independently learns texture patterns, discriminative patches, and shapes to solve various document image analysis tasks. The considered tasks are document image classification, genre identification from book covers, scientific document figure classification, and script identification. The presented network learns global, texture, and discriminative features and combines them judicially based on the nature of the problems to be solved. We compare the performance of the proposed approach with state-of-the-art techniques on multiple publicly available datasets such as Book-Cover, rvl-cdip, cvsi and docfigure. Experiments show that our approach outperforms state-of-the-art for the genre and document figure classifications and obtains comparable results for document image and script classification tasks. © 2022, The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd.

关键词： Document analysis Document figure classification Document image classification Identification of the book’s genre Script identification Texture feature

来源：评论

学校读者我要写书评

暂无评论

LineTR: Unified Text Line Segmentation for Challenging Palm Leaf Manuscripts 27th

LineTR: Unified Text Line Segmentation for Challenging Palm...

引用

27th international Conference on Pattern Recognition, ICPR 2024

作者： Agrawal, Vaibhav Vadlamudi, Niharika Waseem, Muhammad Joseph, Amal Chitluri, Sreenya Sarvadevabhatla, Ravi Kiran Centre for Visual Information Technology International Institute of Information Technology Hyderabad Hyderabad500032 India

ISBN: (纸本)9783031784941

The dense and unstructured text in historical manuscripts presents significant challenges for precise line segmentation due to large diversity in sizes, scripts and appearances of the documents. Existing approaches tackle this complexity either by performing dataset-specific processing or training per-dataset models. This strategy hampers maintainability and scalability as newer manuscript collections get digitized and annotated. In this paper, we propose LineTR, a novel two-stage line segmentation approach which can process a diverse variety of challenging handwritten documents in a unified, dataset-agnostic manner. LineTR’s first stage processes context-adaptive image patches. It consists of a novel DETR-style network which generates parametric representations of text strike-through lines (scribbles) and a novel hybrid CNN-transformer network which generates a text energy map. A dataset-agnostic and robust post-processing procedure is applied on first-stage outputs to obtain document-level scribbles. In the second stage, these scribbles and the text energy map are used within a seam generation framework to obtain highly precise polygons enclosing the manuscript text lines. We also introduce three new diverse text line segmentation datasets comprising challenging Indic and South-East Asian manuscripts. Through experiments, ablations and evaluations, we show that LineTR generates significantly superior line segmentations - all with a single model. Our results also highlight the effectiveness of our unified model for good quality zero-shot inference on the newly introduced datasets. Project page: https://***/LineTR/. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Supervision by Landmarks: An Enhanced Facial De-occlusion Network for VR-Based Applications 17th

Supervision by Landmarks: An Enhanced Facial De-occlusion N...

引用

17th European Conference on Computer Vision, ECCV 2022

作者： Gupta, Surabhi Jinka, Sai Sagar Sharma, Avinash Namboodiri, Anoop Center for Visual Information Technology International Institute of Information Technology Hyderabad India

ISBN: (纸本)9783031250712

Face possesses a rich spatial structure that can provide valuable cues to guide various face-related tasks. The eyes are considered an important socio-visual cue for effective communication. They are an integral feature of facial expressions as they are an important aspect of interpersonal communication. However, virtual reality headsets occlude a significant portion of the face and restrict the visibility of certain facial features, particularly the eye region. Reproducing this region with realistic content and handling complex eye movements such as blinks is challenging. Previous facial inpainting methods are not capable enough to capture subtle eye movements. In view of this, we propose a working solution to refine the reconstructions, particularly around the eye region, by leveraging inherent eye structure. We introduce spatial supervision and a novel landmark predictor module to regularize per-frame reconstructions obtained from an existing image-based facial de-occlusion network. Experiments verify the usefulness of our approach in enhancing the quality of reconstructions to capture subtle eye movements. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Virtual reality

来源：评论

学校读者我要写书评

暂无评论

Self-Supervised 3D Mesh Object Retrieval 14

Self-Supervised 3D Mesh Object Retrieval

引用

14th Indian Conference on Computer Vision, Graphics and Image Processing, ICVGIP 2023

作者： Sanklecha, Kajal Mathur, Prayushi Narayanan, P.J. Center for Visual Information Technology International Institute of Information Technology Hyderabad India

ISBN: (纸本)9798400716256

Digital representations of 3D objects are increasingly being used for engineering, entertainment, education, etc. Efforts to search and retrieve digital 3D models from a collection have not attracted sufficient attention, unlike digital representations of documents, images, etc. Supervised methods are not feasible to solve this problem as a large collection of labelled 3D objects is difficult to create. This paper presents a self-supervised method to learn efficient embeddings of 3D mesh objects for ranked retrieval of similar objects. We propose a simple representation of mesh objects and an encoder-decoder architecture to learn the embedding. Extensive experiments show that our method is competitive with methods that need supervision while being more scalable to different object collections. © 2023 Copyright held by the owner/author(s). Publication rights licensed to ACM.

关键词： Embeddings

来源：评论

学校读者我要写书评

暂无评论

Attribute Reduction on Decision Tables Based on Hausdorff Topology

引用

Computers, Materials & Continua 2024年第11期81卷 3097-3124页

作者： Nguyen Long Giang Tran Thanh Dai Le Hoang Son Tran Thi Ngan Nguyen Nhu Son Cu Nguyen Giap Institute of Information Technology Vietnam Academy of Science and TechnologyHanoi10000Vietnam Faculty of Information Technology University of Economics-Technology for IndustriesHanoi10000Vietnam VNU Information Technology Institute Vietnam National UniversityHanoi10000Vietnam International School Vietnam National UniversityHanoi10000Vietnam Center of Science and Technology Research and Development Thuongmai UniversityHanoi10000Vietnam

Attribute reduction through the combined approach of Rough Sets(RS)and algebraic topology is an open research topic with significant potential for *** research works have introduced a strong relationship between RS and topology spaces for the attribute reduction ***,the mentioned recent methods followed a strategy to construct a new measure for attribute ***,the strategy for searching for the reduct is still to select each attribute and gradually add it to the ***,those methods tended to be inefficient for high-dimensional *** overcome these challenges,we use the separability property of Hausdorff topology to quickly identify distinguishable attributes,this approach significantly reduces the time for the attribute filtering stage of the *** addition,we propose the concept of Hausdorff topological homomorphism to construct candidate reducts,this method significantly reduces the number of candidate reducts for the wrapper stage of the *** are the two main stages that have the most effect on reducing computing time for the attribute reduction of the proposed algorithm,which we call the Cluster Filter Wrapper algorithm based on Hausdorff *** validation on the UCI Machine Learning Repository Data shows that the proposed method achieves efficiency in both the execution time and the size of the reduct.

关键词： Hausdorff topology rough sets topology from rough sets attribute reduction

来源：评论

学校读者我要写书评

暂无评论

Lupus Nephritis Subtype Classification with only Slide Level labels 7

Lupus Nephritis Subtype Classification with only Slide Level...

引用

7th international Conference on Medical Imaging with Deep Learning, MIDL 2024

作者： Sharma, Amit Chauhan, Ekansh Uppin, Megha S. Rajasekhar, Liza Jawahar, C.V. Vinod, P.K. Center for Visual Information Technology International Institute of Information Technology Hyderabad India Department of Pathology Nizam’s Institute of Medical Sciences Hyderabad India Department of Clinical Immunology and Rheumatology Nizam’s Institute of Medical Sciences Hyderabad India Center for Computational Natural Sciences and Bioinformatics International Institute of Information Technology Hyderabad India

Lupus Nephritis classification has historically relied on labor-intensive and meticulous glomerular-level labeling of renal structures in whole slide images (WSIs). However, this approach presents a formidable challenge due to its tedious and resource-intensive nature, limiting its scalability and practicality in clinical settings. In response to this challenge, our work introduces a novel methodology that utilizes only slide-level labels, eliminating the need for granular glomerular-level labeling. A comprehensive multi-stained lupus nephritis digital histopathology WSI dataset was created from the Indian population, which is the largest of its kind. LupusNet, a deep learning MIL-based model, was developed to classify LN subtypes. The results underscore its effectiveness, achieving an AUC score of 91.0%, an F1 score of 77.3%, and an accuracy of 81.1% on our dataset in distinguishing membranous and diffused classes of LN. © 2024 CC-BY 4.0, A. Sharma, E. Chauhan, M.S. Uppin, L. Rajasekhar, C.V. Jawahar & P.K. Vinod.

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

LLM Driven Web Profile Extraction for Identical Names 24

LLM Driven Web Profile Extraction for Identical Names

引用

33rd Companion of the ACM World Wide Web Conference, WWW 2023

作者： Sancheti, Prateek Karlapalem, Kamalakar Vemuri, Kavita Data Sciences and Analytics Center International Institute of Information Technology Hyderabad India Cognitive Science Lab International Institute of Information Technology Hyderabad India

ISBN: (纸本)9798400701726

The number of individuals having identical names on the internet is increasing. Thus making the task of searching for a specific individual tedious. The user must vet through many profiles with identical names to get to the actual individual of interest. The online presence of an individual forms the profile of the individual. We need a solution that helps users by consolidating the profiles of such individuals by retrieving factual information available on the web and providing the same as a single result. We present a novel solution that retrieves web profiles belonging to those bearing identical Full Names through an end-to-end pipeline. Our solution involves information retrieval from the web (extraction), LLM-driven Named Entity Extraction (retrieval), and standardization of facts using Wikipedia, which returns profiles with fourteen multi-valued attributes. After that, profiles that correspond to the same real-world individuals are determined. We accomplish this by identifying similarities among profiles based on the extracted facts using a Prefix Tree inspired data structure (validation) and utilizing ChatGPT’s contextual comprehension (revalidation). The system offers varied levels of strictness while consolidating these profiles, namely strict, relaxed, and loose matching. The novelty of our solution lies in the innovative use of GPT – a highly powerful yet an unpredictable tool, for such a nuanced task. A study involving twenty participants, along with other results, found that one could effectively retrieve information for a specific individual. © 2024 Copyright held by the owner/author(s). Publication rights licensed to ACM.

关键词： Extraction

来源：评论

学校读者我要写书评

暂无评论

Self-Supervised 3D Mesh Object Retrieval 23

Self-Supervised 3D Mesh Object Retrieval

引用

Proceedings of the Fourteenth Indian Conference on Computer Vision, Graphics and Image Processing

作者： Kajal Sanklecha Prayushi Mathur P. J. Narayanan Center of Visual Information Technology International Institute of Information Technology Hyderabad IN Center of Visual Information Technology IIIT-Hyderabad IN

ISBN: (纸本)9798400716256

关键词： 3D Triangle Mesh Embedding Space Mesh Analysis Object Retrieval Ranked Retrieval Self-Supervision

来源：评论

学校读者我要写书评

暂无评论

Multiobjective Differential Evolution for Higher-Dimensional Multimodal Multiobjective Optimization

引用

IEEE/CAA Journal of Automatica Sinica 2024年第6期11卷 1458-1475页

作者： Jing Liang Hongyu Lin Caitong Yue Ponnuthurai Nagaratnam Suganthan Yaonan Wang School of Electrical and Information Engineering Zhengzhou UniversityZhengzhou 450001 School of Electrical Engineering and Automation Henan Institute of TechnologyXinxiang 453003China School of Electrical and Information Engineering Zhengzhou UniversityZhengzhou 450001China KINDI Center for Computing Research Qatar UniversityDohaQatarand also with the Nanyang Technology UniversitySingapore College of Electrical and Information Engineering and also with the National Engineering Research Center for Robot Visual Perception and Control TechnologyHunan UniversityChangsha 410082China IEEE

In multimodal multiobjective optimization problems(MMOPs),there are several Pareto optimal solutions corre-sponding to the identical objective *** paper proposes a new differential evolution algorithm to solve MMOPs with higher-dimensional decision *** to the increase in the dimensions of decision variables in real-world MMOPs,it is diffi-cult for current multimodal multiobjective optimization evolu-tionary algorithms(MMOEAs)to find multiple Pareto optimal *** proposed algorithm adopts a dual-population framework and an improved environmental selection *** utilizes a convergence archive to help the first population improve the quality of *** improved environmental selection method enables the other population to search the remaining decision space and reserve more Pareto optimal solutions through the information of the first *** combination of these two strategies helps to effectively balance and enhance conver-gence and diversity *** addition,to study the per-formance of the proposed algorithm,a novel set of multimodal multiobjective optimization test functions with extensible decision variables is *** proposed MMOEA is certified to be effective through comparison with six state-of-the-art MMOEAs on the test functions.

关键词： Benchmark functions diversity measure evolution-ary algorithms multimodal multiobjective optimization.

来源：评论

学校读者我要写书评

暂无评论

Sparsity-based Channel Estimation for RIS-aided MIMO mmWave Communication Systems 35

Sparsity-based Channel Estimation for RIS-aided MIMO mmWave ...

引用

35th IEEE international Symposium on Personal, Indoor and Mobile Radio Communications, PIMRC 2024

作者： Motwani, Yash Kota, Kali Krishna Mankar, Praful D. International Institute of Information Technology Signal Processing and Communication Research Center Hyderabad India

ISBN: (纸本)9798350362244

Reconfigurable intelligent surfaces (RIS) are emerging as a promising technology for 6 G networks due to their ability to shape the radio propagation environment. This ability helps to overcome propagation challenges, such as high path loss, absorption loss, etc., that are particularly needed at millimetre wave (mmWave) bands. Thus, RIS can enhance the capacity of next-generation mmWave communication networks. However, one critical aspect in the design of RIS-aided systems is channel estimation, as it involves estimating two channels (i.e., base stationRIS and RIS-user terminal) separately based on the compound channel observed by the receiver. In this paper, we address this problem by proposing an algorithm that estimates both channels separately by leveraging the advantages of sparsity of the mmWave channel. Specifically, the proposed algorithm uses the parallel factor (PARAFAC) decomposition of the tensor, which is constructed using the received signal matrices observed under different RIS phase shift configurations. Our numerical analysis shows that the proposed algorithm provides significantly smaller normalized mean square error (NMSE) than the widely used alternating least squares (ALS) algorithm, particularly when the number of RIS phase shift configurations is smaller than the number of RIS elements. © 2024 IEEE.

关键词： Mean square error

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：