检索结果-内蒙古大学图书馆

SSRN 2024年

作者： Cordeiro, Filipe R. Carneiro, Gustavo Visual Computing Lab Department of Computing Universidade Federal Rural de Pernambuco Brazil Centre for Vision Speech and Signal Processing University of Surrey United Kingdom

An important stage of most state-of-the-art (SOTA) noisy-label learning methods consists of a sample selection procedure that classifies samples from the noisy-label training set into noisy-label or clean-label subsets. The process of sample selection typically consists of one of the two approaches: loss-based sampling, where high-loss samples are considered to have noisy labels, or feature-based sampling, where samples from the same class tend to cluster together in the feature space and noisy-label samples are identified as anomalies within those clusters. Empirically, loss-based sampling is robust to a wide range of noise rates, while feature-based sampling tends to work effectively in particular scenarios, e.g., the filtering of noisy instances via their eigenvectors (FINE) sampling exhibits greater robustness in scenarios with low noise rates, and the K nearest neighbor (KNN) sampling mitigates better high noise-rate problems. This paper introduces the Adaptive Nearest Neighbors and Eigenvector-based (ANNE) sample selection methodology, a novel approach that integrates loss-based sampling with the feature-based sampling methods FINE and KNN to optimize performance across a wide range of noise rate scenarios. ANNE achieves this integration by first partitioning the training set into high-loss and low-loss sub-groups using loss-based sampling. Subsequently, within the low-loss subset, sample selection is performed using FINE, while the high-loss subset employs KNN for effective sample selection. We integrate ANNE into noisy-label learning state of the art (SOTA) methods, and test it on CIFAR-10/-100 (with symmetric, asymmetric and instance-dependent noise), © 2024, The Authors. All rights reserved.

关键词： Eigenvalues and eigenfunctions

来源：评论

学校读者我要写书评

暂无评论

InstructIR: High-Quality Image Restoration Following Human Instructions

arXiv

引用

arXiv 2024年

作者： Conde, Marcos V. Geigle, Gregor Timofte, Radu Computer Vision Lab CAIDAS IFI University of Würzburg Germany Visual Computing Group FTG Sony PlayStation

Image restoration is a fundamental problem that involves recovering a high-quality clean image from its degraded observation. All-In-One image restoration models can effectively restore images from various types and levels of degradation using degradation-specific information as prompts to guide the restoration model. In this work, we present the first approach that uses human-written instructions to guide the image restoration model. Given natural language prompts, our model can recover high-quality images from their degraded counterparts, considering multiple degradation types. Our method, InstructIR, achieves state-of-the-art results on several restoration tasks including image denoising, deraining, deblurring, dehazing, and (low-light) image enhancement. InstructIR improves +1dB over previous all-in-one restoration methods. Moreover, our dataset and results represent a novel benchmark for new research on text-guided image restoration and enhancement. © 2024, CC BY-NC-ND.

关键词： Restoration

来源：评论

学校读者我要写书评

暂无评论

INRetouch: Context Aware Implicit Neural Representation for Photography Retouching

arXiv

引用

arXiv 2024年

作者： Elezabi, Omar Conde, Marcos V. Wu, Zongwei Timofte, Radu Computer Vision Lab CAIDAS & IFI University of Würzburg Germany Visual Computing Group FTG Sony PlayStation

Professional photo editing remains challenging, requiring extensive knowledge of imaging pipelines and significant expertise. With the ubiquity of smartphone photography, there is an increasing demand for accessible yet sophisticated image editing solutions. While recent deep learning approaches, particularly style transfer methods, have attempted to automate this process, they often struggle with output fidelity, editing control, and complex retouching capabilities. We propose a novel retouch transfer approach that learns from professional edits through before-after image pairs, enabling precise replication of complex editing operations. To facilitate this research direction, we introduce a comprehensive Photo Retouching Dataset comprising 100,000 high-quality images edited using over 170 professional Adobe Lightroom presets. We develop a context-aware Implicit Neural Representation that learns to apply edits adaptively based on image content and context, requiring no pretraining and capable of learning from a single example. Our method extracts implicit transformations from reference edits and adaptively applies them to new images. Through extensive evaluation, we demonstrate that our approach not only surpasses existing methods in photo retouching but also enhances performance in related image reconstruction tasks like Gamut Mapping and Raw Reconstruction. By bridging the gap between professional editing capabilities and automated solutions, our work presents a significant step toward making sophisticated photo editing more accessible while maintaining high-fidelity results. © 2024, CC BY.

关键词： Photomapping

来源：评论

学校读者我要写书评

暂无评论

End-to-End Learnable Multi-Scale Feature Compression for VCM

arXiv

引用

arXiv 2023年

作者： Kim, Yeongwoong Jeong, Hyewon Yu, Janghyun Kim, Younhee Lee, Jooyoung Jeong, Se Yoon Kim, Hui Yong Visual Media Lab School of Computing Kyung Hee University Korea Republic of Korea Republic of

The proliferation of deep learning-based machine vision applications has given rise to a new type of compression, so called video coding for machine (VCM). VCM differs from traditional video coding in that it is optimized for machine vision performance instead of human visual quality. In the feature compression track of MPEG-VCM, multi-scale features extracted from images are subject to compression. Recent feature compression works have demonstrated that the versatile video coding (VVC) standard-based approach can achieve a BD-rate reduction of up to 96% against MPEG-VCM feature anchor. However, it is still sub-optimal as VVC was not designed for extracted features but for natural images. Moreover, the high encoding complexity of VVC makes it difficult to design a lightweight encoder without sacrificing performance. To address these challenges, we propose a novel multi-scale feature compression method that enables both the end-to-end optimization on the extracted features and the design of lightweight encoders. The proposed model combines a learnable compressor with a multi-scale feature fusion network so that the redundancy in the multi-scale features is effectively removed. Instead of simply cascading the fusion network and the compression network, we integrate the fusion and encoding processes in an interleaved way. Our model first encodes a larger-scale feature to obtain a latent representation and then fuses the latent with a smaller-scale feature. This process is successively performed until the smallest-scale feature is fused and then the encoded latent at the final stage is entropy-coded for transmission. The results show that our model outperforms previous approaches by at least 52% BD-rate reduction and has ×5 to ×27 times less encoding time for object detection. It is noteworthy that our model can attain near-lossless task performance with only 0.002-0.003% of the uncompressed feature data size. Copyright © 2023, The Authors. All rights reserved.

关键词： Image compression

来源：评论

学校读者我要写书评

暂无评论

STREAMING NEURAL IMAGES

arXiv

引用

arXiv 2024年

作者： Conde, Marcos V. Bigos, Andy Timofte, Radu Computer Vision Lab CAIDAS & IFI University of Würzburg Germany Visual Computing Group FTG Sony PlayStation Japan

Implicit Neural Representations (INRs) are a novel paradigm for signal representation that have attracted considerable interest for image compression. INRs offer unprecedented advantages in signal resolution and memory efficiency, enabling new possibilities for compression techniques. However, the existing limitations of INRs for image compression have not been sufficiently addressed in the literature. In this work, we explore the critical yet overlooked limiting factors of INRs, such as computational cost, unstable performance, and robustness. Through extensive experiments and empirical analysis, we provide a deeper and more nuanced understanding of implicit neural image compression methods such as Fourier Feature Networks and Siren. Our work also offers valuable insights for future research in this area. © 2024, CC BY-NC-ND.

关键词： Image compression

来源：评论

学校读者我要写书评

暂无评论

Improving Mass Detection in Mammography Images: A Study of Weakly Supervised Learning and Class Activation Map Methods

arXiv

引用

arXiv 2023年

作者： Sampaio, Vicente Cordeiro, Filipe R. Department of Computing Universidade Federal Rural de Pernambuco Recife Brazil Visual Computing Lab Department of Computing Universidade Federal Rural de Pernambuco Recife Brazil

In recent years, weakly supervised models have aided in mass detection using mammography images, decreasing the need for pixel-level annotations. However, most existing models in the literature rely on Class Activation Maps (CAM) as the activation method, overlooking the potential benefits of exploring other activation techniques. This work presents a study that explores and compares different activation maps in conjunction with state-of-the-art methods for weakly supervised training in mammography images. Specifically, we investigate CAM, GradCAM, GradCAM++, XGradCAM, and LayerCAM methods within the framework of the GMIC model for mass detection in mammography images. The evaluation is conducted on the VinDr-Mammo dataset, utilizing the metrics Accuracy, True Positive Rate (TPR), False Negative Rate (FNR), and False Positive Per Image (FPPI). Results show that using different strategies of activation maps during training and test stages leads to an improvement of the model. With this strategy, we improve the results of the GMIC method, decreasing the FPPI value and increasing TPR. © 2023, CC BY.

关键词： Chemical activation

来源：评论

学校读者我要写书评

暂无评论

Improving Mass Detection in Mammography Images: A Study of Weakly Supervised Learning and Class Activation Map Methods

Improving Mass Detection in Mammography Images: A Study of W...

引用

Brazilian Symposium on Computer Graphics and Image Processing (SIBGRAPI)

作者： Vicente Sampaio Filipe R. Cordeiro Department of Computing Universidade Federal Rural de Pernambuco Recife Brazil Department of Computing Visual Computing Lab Universidade Federal Rural de Pernambuco Recife Brazil

关键词：

来源：评论

学校读者我要写书评

暂无评论

Voice Emotion Recognition Based on Color Histogram Features

Voice Emotion Recognition Based on Color Histogram Features

引用

Annual IEEE Symposium on Computer-Based Medical Systems

作者： Marcelo Marques da Rocha Aura Conci Débora Christina Muchaluat Saade MídiaCom Lab Fluminense Federal University Institute of Computing Niterói Brazil Visual Lab Fluminense Federal University Institute of Computing Niterói Brazil

Voice is the fastest and most efficient method of communication among humans. Researchers believe that it can also be considered the most efficient means of communication between humans and machines. Voice can, in addition to providing useful information, inform us about the emotional state of the person who is speaking. For many applications, being able to identify the emotion is crucial, as it allows the application to adapt to the user. In the case of human-robot interaction, recognizing the user's emotion allows the robot to be more empathetic during interactions. In addition to other methods, such as recognition of facial expressions and recognition through body expressions, recognition of emotions through speech can be used as an additional component in identification the user's emotional state. This work proposes the recognition of emotion through speech using an approach based on image processing of the voice audio signal spectrogram. Two new features based on color histograms are proposed. One thousand six hundred audio files with phrases considering four types of emotions (angry, happy, neutral and sad) were processed and classified. These phrases were spoken by women half by a 64 years old (Subject- 64) and the rest by a 26 years old one (Subject-26). These files are a subset of the TESS (Toronto Emotional Speech Set) dataset. When processing subject-26's voice, an precision of 94.40 % and 91.90 % was achieved in detecting neutral and sad emotions, respectively. When processing subject-64's voice, an precision of 97.00 % was achieved for the angry emotion. The results obtained show the proposal great potential.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Concurrent Charging with Wave Interference 42

Concurrent Charging with Wave Interference

引用

42nd IEEE International Conference on Computer Communications, INFOCOM 2023

作者： Ma, Yuzhuo Wu, Die Ren, Meixuan Peng, Jian Yang, Jilin Liu, Tang Sichuan Normal University College of Computer Science Chengdu610101 China Sichuan Normal University Visual Computing and Virtual Reality Key Lab Chengdu610068 China Sichuan University College of Computer Science Chengdu610065 China

ISBN: (纸本)9798350334142

To improve the charging performance, employing multiple wireless chargers to charge sensors concurrently is an effective way. In such charging scenarios, the radio waves radiated from multiple chargers will interfere with each other. Though a few work have realized the wave interference, they do not fully utilize the high power caused by constructive interference while avoiding the negative impacts brought by the destructive interference. In this paper, we aim to investigate the power distribution regularity of concurrent charging and take full advantage of the high power to enhance the charging efficiency. Specifically, we formulate a concurrent charGing utility mAxImizatioN (GAIN) problem and build a practical charging model with wave interference. Further, we propose a concurrent charging scheme, which not only can improve the power of interference enhanced regions by deploying chargers, but also find a set of points with the highest power to locate sensors. Finally, we conduct both simulations and field experiments to evaluate the proposed scheme. The results demonstrate that our scheme outperforms the comparison algorithms by 40.48% on average. © 2023 IEEE.

关键词： Inductive power transmission

来源：评论

学校读者我要写书评

暂无评论

Sign Language Communication Through an Interactive Mobile Application 25th

Sign Language Communication Through an Interactive Mobile Ap...

引用

25th International Conference on Human-Computer Interaction, HCII 2023

作者： Stergioulas, Andreas Chatzikonstantinou, Christos Chatzis, Theocharis Papastratis, Ilias Konstantinidis, Dimitrios Dimitropoulos, Kosmas Atzakas, Klimnis Xydopoulos, George J. Zacharopoulou, Vassia Papazachariou, Dimitris Aggelidis, Vasilis Grigoriadis, Kostas Daras, Petros Visual Computing Lab Information Technologies Institute Centre for Research and Technology Hellas Thessaloniki Greece University of Patras Patra Greece Centre of Greek Sign Language – KENG Thessaloniki Greece MLS Innovation Inc. Thessaloniki Greece

ISBN: (纸本)9783031359910

Sign languages (SLs) are an essential form of communication for hearing-impaired people. However, a communication barrier still exists between the deaf community and the hearing population due to the lack of accurate automated SL communication systems. In this work, a novel SL communication system running as a mobile application has been developed to facilitate the bi-directional communication between hearing-impaired and hearing *** proposed system utilizes Natural Language Processing (NLP) techniques, along with linguistic rules to convert between spoken language and signs, taking into account the grammatical structure of a sign language. Additionally, the system employs sign language recognition (SLR) algorithms to transform video sequences to signs, as well as hand and pose estimation algorithms to model the 3D motion of signs. Moreover, a 3D human avatar representation is employed to animate the motion of each sign in a seamless manner. Finally, a new partition of the Greek SL (GSL) dataset is formed with 1825 videos from 12 signers captured in the wild to evaluate SLR performance under realistic conditions. The proposed SL communication system and its components are validated quantitatively in GSLW as well as qualitatively by means of questionnaires, demonstrating the user satisfaction with the system. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Audition

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：