检索结果-内蒙古大学图书馆

IMAGE SEGMENTATION FOR IMPROVED LOSSLESS SCREEN CONTENT COMPRESSION

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Uddehal, Shabhrish Reddy Strutz, Tilo Och, Hannah Kaup, André Department of Electrical Engineering and Computer Science Coburg University Coburg Germany Multimedia Communications and Signal Processing Friedrich-Alexander University Erlangen Germany

In recent years, it has been found that screen content images (SCI) can be effectively compressed based on appropriate probability modelling and suitable entropy coding methods such as arithmetic coding. The key objective is determining the best probability distribution for each pixel position. This strategy works particularly well for images with synthetic (textual) content. However, usually screen content images not only consist of synthetic but also pictorial (natural) regions. These images require diverse models of probability distributions to be optimally compressed. One way to achieve this goal is to separate synthetic and natural regions. This paper proposes a segmentation method that identifies natural regions enabling better adaptive treatment. It supplements a compression method known as Soft Context Formation (SCF) and operates as a pre-processing step. If at least one natural segment is found within the SCI, it is split into two sub-images (natural and synthetic parts) and the process of modelling and coding is performed separately for both. For SCIs with natural regions, the proposed method achieves a bit-rate reduction of up to 11.6% and 1.52% with respect to HEVC and the previous version of the SCF. Copyright © 2023, The Authors. All rights reserved.

关键词： Probability distributions

Inappropriate Content Classification based on Video Rating: A Preliminary Study 2

学校读者我要写书评

暂无评论

Inappropriate Content Classification based on Video Rating: ...

2nd International Conference on computer science and Engineering, IC2SE 2021

作者： Yong, Zuo Jun Lam Hoo, Wai University of Malaya Faculty of Computer Science and Information Technology Centre of Image and Signal Processing Department of Information Systems Kuala Lumpur Malaysia

ISBN: (纸本)9781665400459

With the rise of social media, video analysis becomes more important in understanding user behavior. Video analysis has been a challenging problem in the computer vision domain compared to image analysis, considering the number of image frames per second a video can have, and the time-series information that underlying within the video. While recent advancements in video analysis improved gradually, the inappropriate content classification remains challenging, mainly due to the limited number of frames to be detected that is contributing to the inappropriate context of the video. In this work, we tackled this problem by addressing the current algorithm in video analysis problems and investigating the majority algorithm that is widely used in video analysis problems and is heavily affected by the issues in inappropriate content classification. A weighted score strategy is proposed to overcome the limitation of the small number of image frame detection in terms of inappropriate content so that inappropriate content classification problems can be optimized. To prove the concept, a video dataset labeled with inappropriate content inspired by Malaysia movie rating has been collected. The proposed strategy can outperform the majority algorithm in a good margin, showing that the proposed solution helps get more effective in inappropriate content classification. © 2021 IEEE.

关键词： computer vision

Multi-frame Detection via Graph Neural Networks: A Link Prediction Approach

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Lin, Zhihao Gao, Chang Yan, Junkun Zhang, Qingfu Liu, Hongwei National Key Laboratory of Radar Signal Processing Xidian University Xi’an710071 China Department of Computer Science City University of Hong Kong Hong Kong999077 Hong Kong

Multi-frame detection algorithms can effectively utilize the correlation between consecutive echoes to improve the detection performance of weak targets. Existing efficient multiframe detection algorithms are typically based on three sequential steps: plot extraction via a relative low primary threshold, track search and track detection. However, these three-stage processing algorithms may result in a notable loss of detection performance and do not fully leverage the available echo information across frames. As to applying graph neural networks in multi-frame detection, the algorithms are primarily based on node classification tasks, which cannot directly output target tracks. In this paper, we reformulate the multi-frame detection problem as a link prediction task in graphs. First, we perform a rough association of multiframe observations that exceed the low threshold to construct observation association graphs. Subsequently, a multi-feature link prediction network is designed based on graph neural networks, which integrates multi-dimensional information, including echo structure, Doppler information, and spatio-temporal coupling of plots. By leveraging the principle of link prediction, we unifies the processes of track search and track detection into one step to reduce performance loss and directly output target tracks. Experimental results show that, compared with traditional single-frame and multi-frame detection algorithms, the proposed algorithm improves the detection performance of weak targets while suppressing false alarms. Additionally, interpretable analysis indicates that the designed network effectively integrates the utilized features, allowing for accurate associations between targets and false alarms. Copyright © 2024, The Authors. All rights reserved.

关键词： Graph neural networks

Effects of Near-field Sources on Ambisonics Recording and Playback 151

学校读者我要写书评

暂无评论

Effects of Near-field Sources on Ambisonics Recording and Pl...

151st Audio Engineering Society Convention 2021

作者： Gonzalez, Raimundo Politis, Archontis Lokki, Tapio Department of Signal Processing and Acoustics Aalto University Espoo Finland Department of Computer Science Aalto University Espoo Finland Audio and Speech Processing Group Tampere University of Technology Tampere Finland

ISBN: (纸本)9781713839453

Ambisonic recording with spherical microphone arrays (SMAs) is based on a far-field assumption which determines how microphone signals are encoded into Ambisonic signals. In the presence of a near-field source, low-frequency distance-dependent boosts arise in SMAs in similar nature to proximity effects in far-field equalized directional microphones. In this study, the effects of near-field sources on Ambisonic signals are modelled analytically, their interaction with regularization stages is observed, and then traced further across to two basic ambisonic processing operations: virtual microphones, and binaural decoding. © 151st Audio Engineering Society Convention 2021.

关键词： Audio recordings

LoRaWAN and Biosensor-Assisted Smart City Framework to Mitigate the Effects of Lung Diseases

学校读者我要写书评

暂无评论

LoRaWAN and Biosensor-Assisted Smart City Framework to Mitig...

2024 IEEE International Conference on Modeling, Simulation and Intelligent Computing, MoSICom 2024

作者： Abubeker, K.M. Roberts, Michaelraj Kingston Prajitha, C. Rinesh, S. Shafeena Karim, A. Jeevanandham, S. Department of Electronics and Communication Engineering Kerala India Sri Eshwar College of Engineering Centre for Advances in Signal Processing and Wireless Sensor Networks Department of Electronics and Communication Engineering Coimbatore India Karpagam Academy of Higher Education Centre for Interdisciplinary Research Department of Electronics and Communication Engineering Coimbatore India V.S.B College of Engineering Technical Campus Department of Computer Science and Engineering Coimbatore India Sri Ramakrishna Engineering College Department of Information Technology Coimbatore India

ISBN: (纸本)9798331533311

Urbanization and industrialization in cities present substantial obstacles to air quality, resulting in detrimental effects on public health, including respiratory health. This research introduces an Internet of Things (IoT) framework for a smart city environment using LoRaWAN and sensor networks. The framework utilizes the ESP32 LoRaWAN module to provide real-time monitoring of essential air quality metrics, such as carbon monoxide (CO), nitrogen dioxide (NO2), ozone (O3), temperature, and humidity. Integrating biosensors enhances accuracy by identifying distinct biological indicators linked to respiratory well-being. The proposed work aims to use the framework in an urban setting within a global context, demonstrating its capacity to reduce the effects of lung disorders. The ESP32 LoRaWAN module enables effective communication within the sensor network up to 20 kilometers, guaranteeing a robust and adaptable solution for smart city applications. The developed framework provides a scalable and novel method for developing healthier and more sustainable cities, decreasing respiratory disorders as urban settings develop. © 2024 IEEE.

关键词： Lung cancer

Towards the Identifiability in Noisy Label Learning: A Multinomial Mixture Approach

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Nguyen, Cuong Do, Thanh-Toan Carneiro, Gustavo School of Computer Science University of Adelaide Australia Department of Data Science and AI Monash University Australia Centre for Vision Speech and Signal Processing University of Surrey United Kingdom

Learning from noisy labels (LNL) plays a crucial role in deep learning. The most promising LNL methods rely on identifying clean-label samples from a dataset with noisy annotations. Such an identification is challenging because the conventional LNL problem, which assumes a single noisy label per instance, is non-identifiable, i.e., clean labels cannot be estimated theoretically without additional heuristics. In this paper, we aim to formally investigate this identifiability issue using multinomial mixture models to determine the constraints that make the problem identifiable. Specifically, we discover that the LNL problem becomes identifiable if there are at least 2C − 1 noisy labels per instance, where C is the number of classes. To meet this requirement without relying on additional 2C − 2 manual annotations per instance, we propose a method that automatically generates additional noisy labels by estimating the noisy label distribution based on nearest neighbours. These additional noisy labels enable us to apply the Expectation-Maximisation algorithm to estimate the posterior probabilities of clean labels, which are then used to train the model of interest. We empirically demonstrate that our proposed method is capable of estimating clean labels without any heuristics in several label noise benchmarks, including synthetic, web-controlled, and real-world label noises. Furthermore, our method performs competitively with many state-of-the-art methods. © 2023, CC BY.

关键词： Deep learning

Text-Queried Target Sound Event Localization

学校读者我要写书评

暂无评论

Text-Queried Target Sound Event Localization

European signal processing Conference (EUSIPCO)

作者： Jinzheng Zhao Xinyuan Qian Yong Xu Haohe Liu Yin Cao Davide Berghi Wenwu Wang Centre for Vision Speech and Signal Processing (CVSSP) University of Surrey UK Department of Computer Science and Technology University of Science and Technology Beijing China Tencent AI Lab Bellevue WA USA Department of Intelligent Science Xi'an Jiaotong Liverpool University China

ISBN: (数字)9789464593617

ISBN: (纸本)9798331519773

Sound event localization and detection (SELD) aims to determine the appearance of sound classes, together with their Direction of Arrival (DOA). However, current SELD systems can only predict the activities of specific classes, for example, 13 classes in DCASE challenges. In this paper, we propose text- queried target sound event localization (SEL), a new paradigm that allows the user to input the text to describe the sound event, and the SEL model can predict the location of the related sound event. The proposed task presents a more user-friendly way for human-computer interaction. We provide a benchmark study for the proposed task and perform experiments on datasets created by simulated room impulse response (RIR) and real RIR to validate the effectiveness of the proposed methods. We hope that our benchmark will inspire the interest and additional research for text-queried sound source localization.

关键词： Location awareness Human computer interaction Direction-of-arrival estimation Accuracy Europe Benchmark testing signal processing Predictive models

Task Weighting in Meta-learning with Trajectory Optimisation

学校读者我要写书评

暂无评论

arXiv 2023年

Developing meta-learning algorithms that are un-biased toward a subset of training tasks often requires hand-designed criteria to weight tasks, potentially resulting in sub-optimal solutions. In this paper, we introduce a new principled and fully-automated task-weighting algorithm for meta-learning methods. By considering the weights of tasks within the same mini-batch as an action, and the meta-parameter of interest as the system state, we cast the task-weighting meta-learning problem to a trajectory optimisation and employ the iterative linear quadratic regulator to determine the optimal action or weights of tasks. We theoretically show that the proposed algorithm converges to an ∊0-stationary point, and empirically demonstrate that the proposed approach out-performs common hand-engineering weighting methods in two few-shot learning benchmarks. © 2023, CC BY.

关键词： Learning algorithms

A Speaker Recognition Method Based on Stable Learning

学校读者我要写书评

暂无评论

A Speaker Recognition Method Based on Stable Learning

International Conference on Acoustics, Speech, and signal processing (ICASSP)

作者： Jian Zhang Jing Ma Xiaochen Guo Lin Li Liang He School of Computer Science and Technology Xinjiang University China Xinjiang Key Laboratory of Signal Detection and Processing China School of Electronic Science and Engineering Xiamen University China Department of Electronic Engineering Tsinghua University China

With the development of deep learning, speaker recognition systems have shown increasingly better performance. The generalization ability of the models is also an important aspect of performance evaluation. Typically, a baseline system is used to compare against the improved models to demonstrate performance enhancements. However, we cannot determine the differences in learned voiceprint features between the improved models and the baseline system. This paper introduces an improved speaker recognition system based on the ECAPA-TDNN model. It utilizes stable learning to eliminate sample correlation and employs attribution analysis to compare the differences in voiceprint feature learning between the improved and baseline systems. Experimental results demonstrate that stable learning improves the model’s generalization performance and helps it learn better voiceprint features. The effectiveness and generalization capability of the proposed method are verified through experiments on the VoxCeleb, CNCeleb, and LibriSpeech datasets. This work is important for enhancing speaker recognition performance, analyzing differences in voiceprint feature learning, and promoting advancements in the field.

关键词：