检索结果-内蒙古大学图书馆

Opinion Score Distribution Prediction via AI-Based Observers in Media Quality Assessment

学校读者我要写书评

暂无评论

Opinion Score Distribution Prediction via AI-Based Observers...

International Conference on Application of Information and Communication Technologies (AICT)

作者： Lohic Fotio Tiotsop Antonio Servetti Enrico Masala Control and Computer Engineering Department Politecnico di Torino Turin Italy

ISBN: (数字)9798350387537

ISBN: (纸本)9798350387544

Training a Deep Neural Network (DNN) to predict an individual’s opinion score regarding the quality of multimedia content is a recent research direction. This type of DNN is called Artificial Intelligence-based Observer (AIO). By generating individual opinion scores, AIOs enable the prediction of the Opinion Score Distribution (OSD) for a given multimedia content. Multimedia image quality assessment literature lacks contributions that thoroughly assess the ability of AIOs to predict the OSD. In this paper a new set of AIOs is trained and shown to predict the OSD more accurately than state-of-the-art methods.

关键词： Training Image quality Accuracy Artificial neural networks Observers Media Quality assessment Information and communication technology Multimedia communication

METER: a mobile vision transformer architecture for monocular depth estimation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Papa, Lorenzo Russo, Paolo Amerini, Irene the Department of Computer Control and Management Engineering Sapienza University of Rome Italy

Depth estimation is a fundamental knowledge for autonomous systems that need to assess their own state and perceive the surrounding environment. Deep learning algorithms for depth estimation have gained significant interest in recent years, owing to the potential benefits of this methodology in overcoming the limitations of active depth sensing systems. Moreover, due to the low cost and size of monocular cameras, researchers have focused their attention on monocular depth estimation (MDE), which consists in estimating a dense depth map from a single RGB video frame. State of the art MDE models typically rely on vision transformers (ViT) architectures that are highly deep and complex, making them unsuitable for fast inference on devices with hardware constraints. Purposely, in this paper, we address the problem of exploiting ViT in MDE on embedded devices. Those systems are usually characterized by limited memory capabilities and low-power CPU/GPU. We propose METER, a novel lightweight vision transformer architecture capable of achieving state of the art estimations and low latency inference performances on the considered embedded hardwares: NVIDIA Jetson TX1 and NVIDIA Jetson Nano. We provide a solution consisting of three alternative configurations of METER, a novel loss function to balance pixel estimation and reconstruction of image details, and a new data augmentation strategy to improve the overall final predictions. The proposed method outperforms previous lightweight works over the two benchmark datasets: the indoor NYU Depth v2 and the outdoor KITTI. © 2024, CC BY.

关键词： Learning algorithms

Comparative Analysis of Wearable and computer Vision-Based Methods for Human Torso Posture Estimation

学校读者我要写书评

暂无评论

Comparative Analysis of Wearable and Computer Vision-Based M...

International Conference on Mathematics and computers in Sciences and in Industry (MCSI)

作者： Emanuel Muntean Monica Leba Department of System Control and Computer Engineering University of Petrosani Petrosani Romania

ISBN: (数字)9798350377224

ISBN: (纸本)9798350377231

This paper conducts a comparative analysis of human torso posture estimation methodologies, focusing on an inertial measurement unit (IMU) sensor coupled with an Arduino UNO as a wearable approach, and Kinect V2, utilizing OpenCV for posture analysis. The core objective of this study is to ascertain which method yields greater precision in the estimation of human torso posture. The IMU sensor, characterized by its wearability, provides the distinct advantage of unobtrusiveness and the capability to record motion across diverse environments. Conversely, Kinect V2 leverages computer vision techniques to derive posture estimations from visual data in real-time. Through comprehensive experimentation, this research evaluates the accuracy of both methodologies by juxtaposing their posture estimation outcomes. The findings of this investigation aim to significantly contribute to the enhancement of human posture estimation systems, with wide-ranging implications for health, sports, and ergonomics.

关键词： Torso Visualization Accuracy Measurement units Estimation Inertial navigation Sensor fusion Reliability Wearable devices Sports

The Quest for Believability: Exploring FACS Adaptations for Emotion Facial Expressions in Virtual Humans

学校读者我要写书评

暂无评论

The Quest for Believability: Exploring FACS Adaptations for ...

IEEE Games Media Entertainment (GEM)

作者： Stefano Calzolari Francesco Strada Andrea Bottino Control and Computer Engineering Department Politecnico di Torino Torino Italy

ISBN: (数字)9798350374537

ISBN: (纸本)9798350374544

In interactive computer graphics, Facial Action Coding System (FACS) has been adapted to enhance the emotional expressiveness of Virtual Humans (VHs) by associating certain Action Units (AUs) with corresponding facial blendshapes. In this way, animators can (theoretically) recreate any human emotion on a VH's face with precision and flexibility. However, conveying realistic and believable emotional expressions with this approach comes with some challenges. In particular, given a set of AUs representing a particular emotion, it is not straightforward to define the correct set of blendshape weights that can render the same realistic and believable emotion on all VHs, as even small differences in weight values can drastically change the perceived emotion. This complexity raises several critical questions such as: is there for each emotion a universal set of blendshape weights that can effectively convey that emotion across all VHs? How can this set be found? If such a universal set proves elusive, can optimal combinations be identified for specific subgroups based on specific facial features such as men and women? Answering these questions is critical to understanding the general applicability of FACS-based facial emotion coding, which allows designers and animators to easily develop VHs that are able to interact with users in a way that is both emotionally rich and authentic. This paper explores these issues through a preliminary investigation aimed at defining realistic representations of happiness.

关键词： Entertainment industry computer graphics Media Encoding Complexity theory Facial features Faces

ViGEO: an Assessment of Vision GNNs in Earth Observation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Colomba, Luca Garza, Paolo Department of Control and Computer Engineering Politecnico di Torino Turin Italy

Satellite missions and Earth Observation (EO) systems represent fundamental assets for environmental monitoring and the timely identification of catastrophic events, long-term monitoring of both natural resources and human-made assets, such as vegetation, water bodies, forests as well as buildings. Different EO missions enables the collection of information on several spectral bandwidths, such as MODIS, Sentinel-1 and Sentinel-2. Thus, given the recent advances of machine learning, computer vision and the availability of labeled data, researchers demonstrated the feasibility and the precision of land-use monitoring systems and remote sensing image classification through the use of deep neural networks. Such systems may help domain experts and governments in constant environmental monitoring, enabling timely intervention in case of catastrophic events (e.g., forest wildfire in a remote area). Despite the recent advances in the field of computer vision, many works limit their analysis on Convolutional Neural Networks (CNNs) and, more recently, to vision transformers (ViTs). Given the recent successes of Graph Neural Networks (GNNs) on non-graph data, such as time-series and images, we investigate the performances of a recent Vision GNN architecture (ViG) applied to the task of land cover classification. The experimental results show that ViG achieves state-of-the-art performances in multiclass and multilabel classification contexts, surpassing both ViT and ResNet on large-scale benchmarks. Copyright © 2024, The Authors. All rights reserved.

关键词： Image classification

Know Your Bugs: A Survey of Issues in Automated Game Testing Literature

学校读者我要写书评

暂无评论

Know Your Bugs: A Survey of Issues in Automated Game Testing...

IEEE Games Media Entertainment (GEM)

作者： Riccardo Coppola Tommaso Fulcini Francesco Strada Department of Control and Computer Engineering Politecnico di Torino Turin Italy

ISBN: (数字)9798350374537

ISBN: (纸本)9798350374544

As the complexity of video games continues to evolve, so does the importance of effective game testing methodologies. To this end, automated game testing has emerged as a pivotal approach to ensure the quality and functionality of modern games. The objective of the present paper is to identify, through a literature review and the application of Open and Axial coding, the most commonly analysed and mentioned issues in automated game testing literature. The results of the study provide a taxonomy of 26 different issues that are assessed in the software engineering literature by automated game testing practice, grouped in five higher-level categories. The elicited taxonomy can serve as an instrument for game testers, researchers and tool developers to evaluate testing approaches and techniques, enable comparability of research results, and design instruments to investigate functional aspects of games in development.

关键词： Surveys Video games Instruments Taxonomy Entertainment industry Games Media

Enhancing Automotive Embedded Applications: A Comprehensive Evaluation of control Flow Checking Methods

学校读者我要写书评

暂无评论

Enhancing Automotive Embedded Applications: A Comprehensive ...

IEEE International Conference on Design & Test of Integrated Micro & Nano-Systems (DTS)

作者： Mohammadreza Amel Solouki Jacopo Sini Massimo Violante Department of Control and Computer Engineering Politecnico di Torino Turin Italy

ISBN: (数字)9798350363128

ISBN: (纸本)9798350363135

Mitigating the risks posed by Random Hardware Failures (RHFs) is crucial to prevent data corruption and control Flow Errors (CFEs) in embedded systems. This paper addresses these concerns through the application of Software-Implemented Hardware Fault Tolerance (SIHFT) methods, emphasizing compatibility with high-level programming languages such as C. Current SIHFT methods, often implemented in low-level Assembly, present challenges in terms of overhead to code size and real-time execution. Our proposed approach focuses on pre-compilation application of SIHFT methods, specifically control Flow Checking (CFC), to identify CFEs within C-language-based code. We conducted a comparative analysis of two established software-based CFE detection methods in C, seamlessly integrating CFC methods into the application behavioral model. Our methodology ensures ISO26262 compliance, crucial for the automotive sector, offering a software-only strategy that aligns with safety and cost considerations.

关键词： Industries Technical requirements Codes Software design Hardware Software Real-time systems Safety Standards Automotive engineering

Toward a Realistic Benchmark for Out-of-Distribution Detection

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Recalcati, Pietro Garcea, Fabio Piano, Luca Lamberti, Fabrizio Morra, Lia Department of Control and Computer Engineering Politecnico di Torino Torino Italy

Deep neural networks are increasingly used in a wide range of technologies and services, but remain highly susceptible to out-of-distribution (OOD) samples, that is, drawn from a different distribution than the original training set. A common approach to address this issue is to endow deep neural networks with the ability to detect OOD samples. Several benchmarks have been proposed to design and validate OOD detection techniques. However, many of them are based on far-OOD samples drawn from very different distributions, and thus lack the complexity needed to capture the nuances of real-world scenarios. In this work, we introduce a comprehensive benchmark for OOD detection, based on ImageNet and Places365, that assigns individual classes as in-distribution or out-of-distribution depending on the semantic similarity with the training set. Several techniques can be used to determine which classes should be considered in-distribution, yielding benchmarks with varying properties. Experimental results on different OOD detection techniques show how their measured efficacy depends on the selected benchmark and how confidence-based techniques may outperform classifier-based ones on near-OOD samples. Copyright © 2024, The Authors. All rights reserved.

关键词： Deep neural networks

BONES: a Benchmark fOr Neural Estimation of Shapley values

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Napolitano, Davide Cagliero, Luca Department of Control and Computer Engineering Politecnico di Torino Torino Italy

Shapley Values are concepts established for eXplainable AI. They are used to explain black-box predictive models by quantifying the features’ contributions to the model’s outcomes. Since computing the exact Shapley Values is known to be computationally intractable on real-world datasets, neural estimators have emerged as alternative, more scalable approaches to get approximated Shapley Values estimates. However, experiments with neural estimators are currently hard to replicate as algorithm implementations, explainer evaluators and results visualizations are neither standardized nor promptly usable. To bridge this gap, we present BONES, a new benchmark focused on neural estimation of Shapley Value. It provides researchers with a suite of state-of-the-art neural and traditional estimators, a set of commonly used benchmark datasets, ad hoc modules for training black-box models, as well as specific functions to easily compute the most popular evaluation metrics and visualize results. The purpose is to simplify XAI model usage, evaluation, and comparison. In this paper, we showcase BONES results and visualizations for XAI model benchmarking on both tabular and image data. The opensource library is available at the following link: https://***/DavideNapolitano/BONES. © 2024, CC BY.

关键词： Benchmarking