检索结果-内蒙古大学图书馆

41st International Conference on Machine Learning, ICML 2024

作者： Melnyk, Pavlo Felsberg, Michael Wadenbäck, Mårten Robinson, Andreas Le, Cuong Computer Vision Laboratory Department of Electrical Engineering Linköping University Sweden

In this paper, we utilize hyperspheres and regular n-simplexes and propose an approach to learning deep features equivariant under the transformations of nD reflections and rotations, encompassed by the powerful group of O(n). Namely, we propose O(n)-equivariant neurons with spherical decision surfaces that generalize to any dimension n, which we call Deep Equivariant Hyperspheres. We demonstrate how to combine them in a network that directly operates on the basis of the input points and propose an invariant operator based on the relation between two points and a sphere, which as we show, turns out to be a Gram matrix. Using synthetic and real-world data in nD, we experimentally verify our theoretical contributions and find that our approach is superior to the competing methods for O(n)-equivariant benchmark datasets (classification and regression), demonstrating a favorable speed/performance trade-off. The code is available on GitHub. Copyright 2024 by the author(s)

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

MMInstruct: a high-quality multi-modal instruction tuning dataset with extensive diversity

引用

Science China(Information Sciences) 2024年第12期67卷 36-51页

作者： Yangzhou LIU Yue CAO Zhangwei GAO Weiyun WANG Zhe CHEN Wenhai WANG Hao TIAN Lewei LU Xizhou ZHU Tong LU Yu QIAO Jifeng DAI School of Computer Science Nanjing University School of Electronic Information and Electrical Engineering Shanghai Jiao Tong University Shanghai AI Laboratory School of Computer Science Fudan University Department of Information Engineering The Chinese University of Hong Kong SenseTime Research Department of Electronic Engineering Tsinghua University

Despite the effectiveness of vision-language supervised fine-tuning in enhancing the performance of vision large language models(VLLMs), existing visual instruction tuning datasets include the following limitations.(1) Instruction annotation quality: despite existing VLLMs exhibiting strong performance,instructions generated by those advanced VLLMs may still suffer from inaccuracies, such as hallucinations.(2) Instructions and image diversity: the limited range of instruction types and the lack of diversity in image data may impact the model's ability to generate diversified and closer to real-world scenarios outputs. To address these challenges, we construct a high-quality, diverse visual instruction tuning dataset MMInstruct,which consists of 973k instructions from 24 domains. There are four instruction types: judgment, multiplechoice, long visual question answering, and short visual question answering. To construct MMInstruct, we propose an instruction generation data engine that leverages GPT-4V, GPT-3.5, and manual correction. Our instruction generation engine enables semi-automatic, low-cost, and multi-domain instruction generation at 1/6 the cost of manual construction. Through extensive experiment validation and ablation experiments,we demonstrate that MMInstruct could significantly improve the performance of VLLMs, e.g., the model fine-tuning on MMInstruct achieves new state-of-the-art performance on 10 out of 12 benchmarks. The code and data shall be available at https://***/yuecao0119/MMInstruct.

关键词： instruction tuning multi-modal multi-domain dataset vision large language model

来源：评论

学校读者我要写书评

暂无评论

Drowning Recognition for Ocean Surveillance using computer vision and Drone Control

Drowning Recognition for Ocean Surveillance using Computer V...

引用

2023 IEEE International Conference on Electro Information Technology, eIT 2023

作者： Rakotondraibe, Maureen Fang, Tianyang Saniie, Jafar Research Laboratory Department of Electrical and Computer Engineering ChicagoIL United States

ISBN: (纸本)9781665493765

According to WHO's report from 2021, Drowning is the 3rd leading cause of unintentional death worldwide. The use of autonomous drones for drowning recognition can increase the survival rate and help lifeguards and rescuers with their life saving mission. This paper presents a real-time drowning recognition model and algorithm for ocean surveillance that can be implemented on a drone. The presented model has been trained using two different approaches and has 88% accuracy. Compared to the contemporary models of drowning recognition designed for swimming pools, the model presented is better suited for outdoor applications in the ocean. © 2023 IEEE.

关键词： Drones

来源：评论

学校读者我要写书评

暂无评论

Utilizing computer vision Algorithms to Detect and Classify Cyberattacks in IoT Environments in Real-Time

Utilizing Computer Vision Algorithms to Detect and Classify ...

引用

2023 IEEE International Conference on Electro Information Technology, eIT 2023

作者： Gromov, Mikhail Arnold, David Saniie, Jafar Research Laboratory Department of Electrical and Computer Engineering ChicagoIL United States

ISBN: (纸本)9781665493765

computer vision has proven itself capable of accurately detecting and classifying objects within images. This also works in cases where images are used as a way of representing data, without being actual photographs. In cybersecurity, computer vision is rarely used, however it has been used to detect botnets successfully. We applied computer vision to determine how well it would be able to detect and classify a large number of attacks and determined that it would be able to run at a decent rate on a Jetson Nano. This was accomplished by training a convolutional neural network using data publicly available in the IoT-23 database, which contains packet captures of IoT devices with and without different malware infections. The neural network was evaluated on an RTX 3050 and a Jetson Nano to see if it could be used in IoT. © 2023 IEEE.

关键词： computer vision

来源：评论

学校读者我要写书评

暂无评论

IoT-Enabled Smart Bike Helmet with an AI-Driven Collision Avoidance System

IoT-Enabled Smart Bike Helmet with an AI-Driven Collision Av...

引用

2023 IEEE International Conference on Electro Information Technology, eIT 2023

作者： Solus, Jacob Rakotondraibe, Maureen Yu, Xinrui Yi, Won-Jae Gromov, Mikhail Saniie, Jafar Research Laboratory Department of Electrical and Computer Engineering ChicagoIL United States

ISBN: (纸本)9781665493765

This paper presents a system design for a smart bike helmet with multiple safety features that are intended to empower bicycle riders to proactively avoid potential sources of danger or injury. A Smart Sensor/Actuator Node (SSAN), driven by an Arduino Uno single-board microcontroller, contains input sensors and actuators to provide riders the ability to send and receive warnings promptly on their helmet. A vision Node, driven by an NVIDIA Jetson Nano and a cable pin-connected camera, executes AI object detection algorithms for any dangerous objects that are out of sight of the rider and sends alerts to the SSAN as needed. By combining safety features of the SSAN and vision Node while continuously sending data to an IoT-enabled backend web server, the safety operation of a typical bike ride can be substantially improved. © 2023 IEEE.

关键词： Edge computing

来源：评论

学校读者我要写书评

暂无评论

Online Self-distillation and Self-modeling for 3D Brain Tumor Segmentation

引用

IEEE Journal of Biomedical and Health Informatics 2025年第6期29卷 4290-4302页

作者： Pang, Yan Li, Yunhao Huang, Teng Liang, Jiaming Wang, Zhen Dong, Changyu Kuang, Dongyang Hu, Ying Chen, Hao Lei, Tim Wang, Qiong The Guangdong Provincial Key Laboratory of Computer Vision and Virtual Reality Technology Shenzhen Institute of Advanced Technology Chinese Academy of Sciences Shenzhen China The School of Artificial Intelligence Guangzhou University China The Zhejiang Lab Hangzhou China Sun Yat-sen University China The Department of Computer Science and Engineering The Department of Chemical and Biological Engineering Hong Kong University of Science and Technology China The Department of Electrical Engineering University of Colorado Denver United States

In the specialized domain of brain tumor segmentation, supervised segmentation approaches are hindered by the limited availability of high-quality labeled data, a condition arising from data privacy concerns, significant costs, and ethical issues. In response to this challenge, this paper presents a training framework that adeptly integrates a plug-and-play component, MOD, into current supervised learning models, boosting their efficacy in scenarios with limited data. The MOD consists of an Online Tokenizer and a Dense Predictor, which employs self-distillation and self-modeling on masked patches, promoting swift convergence and efficient representation learning. During the inference phase, the plug-and-play MOD component is excluded, preserving the computational efficiency of the original model without incurring extra processing costs. We substantiated the value of our approach through experiments on leading 3D brain tumor segmentation baselines. Remarkably, models augmented with the MOD consistently showcased superior results, achieving elevated Dice coefficients and HD95 scores on two datasets: BraTS 2021 and MSD 2019 Task-01 Brain Tumor. Code: https://***/aigzhusmart/MOD © 2013 IEEE.

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Multi‐granularity re‐ranking for visible‐infrared person re‐identification

引用

CAAI Transactions on Intelligence Technology 2023年第3期8卷 770-779页

作者： Yadi Wang Hongyun Zhang Duoqian Miao Witold Pedrycz Department of Computer Science and Technology Tongji UniversityShanghaiChina Key Laboratory of Embedded System and Service Computing Ministry of EducationShanghaiChina Department of Electrical and Computer Engineering University of AlbertaEdmontonAlbertaCanada System Research Institute Polish Academy of SciencesWarsawPoland

Visible‐infrared person re‐identification(VI‐ReID)is a supplementary task of single‐modality re‐identification,which makes up for the defect of conventional re‐identification under insufficient *** is more challenging than single‐modality ReID because,in addition to difficulties in pedestrian posture,camera shoot-ing angle and background change,there are also difficulties in the cross‐modality *** works only involve coarse‐grained global features in the re‐ranking calculation,which cannot effectively use fine‐grained ***,fine‐grained features are particularly important due to the lack of information in cross‐modality re‐*** this end,the Q‐center Multi‐granularity K‐reciprocal Re‐ranking Algorithm(termed QCMR)is proposed,including a Q‐nearest neighbour centre encoder(termed QNC)and a Multi‐granularity K‐reciprocal Encoder(termed MGK)for a more comprehensive feature *** converts the probe‐corresponding modality features into gallery corresponding modality features through modality transfer to narrow the modality *** takes a coarse‐grained mutual nearest neighbour as the dominant and combines a fine‐grained nearest neighbour as a supplement for similarity *** experiments on two widely used VI‐ReID benchmarks,SYSU‐MM01 and RegDB have shown that our method achieves state‐of‐the‐art ***,the mAP of SYSU‐MM01 is increased by 5.9%in all‐search mode.

关键词： computer vision recognition

来源：评论

学校读者我要写书评

暂无评论

TetraSphere: A Neural Descriptor for O(3)-Invariant Point Cloud Analysis

TetraSphere: A Neural Descriptor for O(3)-Invariant Point Cl...

引用

Conference on computer vision and Pattern Recognition (CVPR)

作者： Pavlo Melnyk Andreas Robinson Michael Felsberg Mårten Wadenbäck Department of Electrical Engineering Computer Vision Laboratory Linköping University Sweden

ISBN: (数字)9798350353006

ISBN: (纸本)9798350353013

In many practical applications, 3D point cloud analy-sis requires rotation invariance. In this paper, we present a learnable descriptor invariant under 3D rotations and reflections, i.e., the O(3) actions, utilizing the recently intro-duced steerable 3D spherical neurons and vector neurons. Specifically, we propose an embedding of the 3D spherical neurons into 4D vector neurons, which leverages end-to-end training of the model. In our approach, we perform TetraTransform-an equivariant embedding of the 3D input into 4D, constructed from the steerable neurons-and ex-tract deeper O(3)-equivariant features using vector neurons. This integration of the TetraTransform into the VN-DGCNN framework, termed TetraSphere, negligibly increases the number of parameters by less than 0.0002%. TetraSphere sets a new state-of-the-art performance classifying randomly rotated real-world object scans of the challenging subsets of ScanObjectNN. Additionally, TetraSphere outperforms all equivariant methods on randomly rotated synthetic data: classifying objects from ModelNet40 and segmenting parts of the ShapeNet shapes. Thus, our results reveal the prac-tical value of steerable 3D spherical neurons for learning in 3D Euclidean space. The code is available at https: //***/pavlo-melnyk/tetrasphere.

关键词： Point cloud compression Training Solid modeling Three-dimensional displays Shape Neurons Vectors

来源：评论

学校读者我要写书评

暂无评论

Machine Learning Safety System for Treadmill Users

Machine Learning Safety System for Treadmill Users

引用

2022 IEEE International Conference on Electro Information Technology, eIT 2022

作者： Natarajan, Ravishankar Yu, Xinrui Saniie, Jafar Department of Electrical and Computer Engineering Research Laboratory ChicagoIL United States

ISBN: (纸本)9781665480093

In this paper, we present an AI-based treadmill assistant that employs computer vision and Deep learning algorithms to offer a comprehensive safety net for treadmill users. The system is designed and implemented with minimal components including an off-the-shelf camera, speaker, an IoT smart plug, and Jetson Nano (a single board GPU-based embedded computing device). To provide an extensive set of safety features such as face recognition, facial stress detection, object detection, and distance and aspect measurement to the treadmill user, the system uses CNN architectures based on ResNet, MobileNet SSD, and customized fully connected neural networks. This set of features monitored in real-time will play a crucial role in mitigating treadmill user falls and injuries. © 2022 IEEE.

关键词： computer vision

来源：评论

学校读者我要写书评

暂无评论

Distributed Gesture Controlled Systems for Human-Machine Interface

Distributed Gesture Controlled Systems for Human-Machine Int...

引用

2022 IEEE International Conference on Electro Information Technology, eIT 2022

作者： Johnson, Hans Saniie, Jafar Department of Electrical and Computer Engineering Research Laboratory ChicagoIL United States

ISBN: (纸本)9781665480093

This paper presents the design flow of an IoT human-machine touchless interface. The device uses embedded computing in conjunction with the Leap Motion Controller to provide an accurate and intuitive touchless interface. Its main function is to augment current touchscreen devices in public spaces through a combination of computer vision technology, event-driven programming, and machine learning. Especially following the COVID-19 pandemic, this technology is important for hygiene and sanitation purposes for public devices such as airport, food, and ATM kiosks where hundreds or even thousands of people may touch these devices in a single day. A prototype of the touchless interface was designed with a Leap Motion Controller housed on a Windows PC exchanging information with a Raspberry Pi microcontroller via internet connection. © 2022 IEEE.

关键词： Internet of things

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：