检索结果-内蒙古大学图书馆

Yolov3-Pruning(transfer): real-time object detection algorithm based on transfer learning

JOURNAL OF REAL-TIME image processing 2022年第4期19卷 839-852页

作者： Li, Xiaoning Wang, Zhengzhong Geng, Shichao Wang, Lin Zhang, Huaxiang Liu, Li Li, Donghua Shandong Normal Univ Sch Informat Sci & Engn Jinan 250014 Shandong Peoples R China Shandong Normal Univ Sch Journalism & Commun Jinan 250014 Shandong Peoples R China Shandong Normal Univ Inst Data Sci & Technol Jinan 250014 Shandong Peoples R China

In recent years, object detection algorithms have achieved great success in the field of machine vision. To pursue the detection accuracy of the model, the scale of the network is constantly increasing, which leads to the continuous increase in computational cost and a large requirement for memory. The larger network scale allows their execution to take a longer time, facing the balance between the detection accuracy and the speed of execution. Therefore, the developed algorithm is not suitable for real-time applications. To improve the detection performance of small targets, we propose a new method, the real-time object detection algorithm based on transfer learning. Based on the baseline Yolov3 model, pruning is done to reduce the scale of the model, and then migration learning is used to ensure the detection accuracy of the model. The object detection method using transfer learning achieves a good balance between detection accuracy and inference speed and is more conducive to the real-time processing of images. Through the evaluation of the dataset voc2007 + 2012, the experimental results show that the parameters of the Yolov3-Pruning(transfer): model are reduced by 3X compared with the baseline Yolov3 model, and the detection accuracy is improved, realizes real-time processing, and improves the detection accuracy.

关键词： Object detection Transfer learning Pruning Detection accuracy Inference speed Real-time processing

来源：评论

学校读者我要写书评

暂无评论

Remote Sensing image Captioning (RSIC): A Technical Review 1st

Remote Sensing Image Captioning (RSIC): A Technical Review

引用

1st International Conference on Data Engineering and machine Intelligence, ICDEMI 2023

作者： Dhinesh, A. Sumathy, P. Department of Computer Science Bharathidasan University Tamilnadu Tiruchirappalli620023 India

ISBN: (纸本)9789819776153

Remote Sensing image Captioning (RSIC) is crucial for many researchers since it has many applications in environmental monitoring, disaster management, urban planning, image retrieval, performance of building planes, military intelligence, and autonomous vehicles. The effective procedure to generate the captions from remote sensing images complements the above-mentioned application domains. Various baseline data sets have been created by the researchers to enhance the quality of captioning by processing the diverse features of the geospatial information. In this paper, we have technically reviewed important literature that follow different algorithms for generating the captions. For example, we have presented the technical review on vision-Language Aligning Paradigm (VLCA) under the bi-lingual caption generation model, Joint-Training Two-Stage (JTTS) technique under multimodel fusion category, Multilevel and Contextual Attention Network (MLCA-Net) under context-aware captioning, LEVIR-CC belongs to transfer learning model, BERT and GPT-3 models belong to transfer-based model, Multiscale Attention (MSA) and Multifeat Attention (MFA) of Multiscale captioning model and Summarization Driven (SD)-RSIC of fine-grained captioning model. We have also presented the performance of each of these methods on various benchmark datasets. For evaluation, different well-known performance metrics are considered. The result is critically evaluated and commented on. In the future, a more rigorous review of these methods along with other relevant methods will be presented along with implementation data. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Urban planning

来源：评论

学校读者我要写书评

暂无评论

Real-Time Object Detection and Tracking Design Using Deep Learning with Spatial–Temporal Mechanism for Video Surveillance applications 10th

Real-Time Object Detection and Tracking Design Using Deep Le...

引用

10th International Conference on Innovations in Computer Science and Engineering, ICICSE 2022

作者： Kusuma, T. Ashwini, K. Global Academy of Technology Bangalore India

ISBN: (纸本)9789811974540

We propose a CNN-based framework for "real-time object detection and tracking using deep learning" in this paper, which includes a spatial–temporal mechanism. The impact of efficient data on performance benchmarks in terms of accuracy has changed. The data processing is handled by industry buzzwords: deep learning (DL) and computer vision (CV). The CNN-based framework uses the single object tracker value to match arrival models and find targets in the next frame. Simply applying single object tracking to multiple object tracking will encounter problems in computational efficiency and results due to occlusion. In this paper, we introduce a "spatial attention mechanism (STAM)" to manage occlusion bias and target interaction. Object tracking is a sensational technology in image processing with great future implications. Multiple object tracking (MOT) has seen an extensive boom in the last few years due to machine learning, deep learning, computer vision, and more. This paper aims to provide an object tracking software solution. Using YOLO’s "You Only Look Once" technology with the help of Tensor flow, the system is geared toward object detection, tracking, and counting. Proven, effective detection and tracking on various dataset. Algorithms that offer real-time, accurate, and precise identifications appropriate for real-time applications. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

Advancements in Structural Health Monitoring Using Combined Computer-vision and Unmanned Aerial Vehicles Approaches 10th

Advancements in Structural Health Monitoring Using Combined ...

引用

10th European Workshop on Structural Health Monitoring (EWSHM)

作者： Sabato, Alessandro Niezrecki, Christopher Dabetwar, Shweta Kulkarni, Nitin Nagesh Bottalico, Fabio Nieduzak, Tymon Univ Massachusetts Lowell Dept Mech Engn Lowell MA 01852 USA

ISBN: (纸本)9783031072581;9783031072574

Aerospace, civil, energy, and mechanical engineering structures continue to be used despite reaching their design lifetime. Developing sensing and data analytics to assess the structural condition of the targeted systems is crucial. Traditional contact-based techniques may produce inconsistent results and are labor-intensive to be considered a valid alternative for monitoring large-scale structures such as bridges, large buildings, and wind turbines. Advancements in image-processing algorithms made techniques such as three-dimensional digital image correlation (3D-DIC), infrared thermography (IRT), motion magnification (MM), and structure from motion (SfM) appealing tools for structural health monitoring and non-destructive testing. Besides, as those techniques are implemented within unmanned aerial vehicles (UAVs), the measurement process is expedited while reducing interference with the targeted structure. This paper summarizes the research experience performed at the University of Massachusetts Lowell. The results of these activities show that the combination of autonomous flight with 3D-DIC, IRT, and SfM can provide precious insights into the structural conditions of the inspected systems while reducing downtime and costs. The study includes future research directions to make those approaches suitable for real-world applications.

关键词： Computer vision Digital image correlation Infrared imaging Motion magnification Optical techniques Structure from motion Unmanned aerial vehicles

来源：评论

学校读者我要写书评

暂无评论

Exploring photosensitive nanomaterials and optoelectronic synapses for neuromorphic artificial vision

引用

CURRENT OPINION IN SOLID STATE & MATERIALS SCIENCE 2025年 35卷

作者： Lee, Hyun-Haeng Ro, Jun-Seok Kim, Kwan-Nyeong Park, Hea-Lim Lee, Tae-Woo Seoul Natl Univ Dept Mat Sci & Engn Seoul 08826 South Korea Seoul Natl Univ Sci & Technol Dept Mat Sci & Engn Seoul 01811 South Korea Seoul Natl Univ Inst Engn Res Res Inst Adv Mat Dept Chem & Biol EngnInterdisciplinary Program Bi 1 Gwanak Ro Seoul 08826 South Korea SN Display Co Ltd Seoul 08826 South Korea

Artificial vision systems will be essential in intelligent machine-vision applications such as autonomous vehicles, bionic eyes, and humanoid robot eyes. However, conventional digital electronics in these systems face limitations in system complexity, processing speed, and energy consumption. These challenges have been addressed by biomimetic approaches utilizing optoelectronic synapses inspired by the biological synapses in the eye. Nano- materials can confine photogenerated charge carriers within nano-sized regions, and thus offer significant potential for optoelectronic synapses to perform in-sensor image-processing tasks, such as classifying static multicolor images and detecting dynamic object movements. We introduce recent developments in optoelectronic synapses, focusing on use of photosensitive nanomaterials. We also explore applications of these synapses in recognizing static and dynamic optical information. Finally, we suggest future directions for research on optoelectronic synapses to implement neuromorphic artificial vision.

关键词： Optoelectronic synapses Nanomaterials Artificial vision systems Artificial synapses Neuromorphic bioelectronics

来源：评论

学校读者我要写书评

暂无评论

Advancements in low light image enhancement techniques and recent applications

引用

JOURNAL OF VISUAL COMMUNICATION AND image REPRESENTATION 2024年 103卷

作者： Anoop, P. P. Deivanathan, R. Vellore Inst Technol Sch Mech Engn Chennai 600127 India

Low-light image enhancement is an effective solution for improving image recognition by both humans and machines. Due to low illuminance, images captured in such conditions possess less color information compared to those taken in daylight, resulting in occluded images characterized by distortion, low contrast, low brightness, a narrow gray range, and noise. Low-light image enhancement techniques play a crucial role in enhancing the effectiveness of object detection. This paper reviews state-of-the-art low-light image enhancement techniques and their developments in recent years. Techniques such as gray transformation, histogram equalization, defogging, Retinex, image fusion, and wavelet transformation are examined, focusing on their working principles and assessing their ability to improve image quality. Further discussion addresses the contributions of deep learning and cognitive approaches, including attention mechanisms and adversarial methods, to image enhancement.

关键词： Computer vision Low-light image enhancement Deep learning image processing machine learning image quality assessment

来源：评论

学校读者我要写书评

暂无评论

Towards machine learning for heterogeneous inverse scattering in 3D microscopy

引用

OPTICS EXPRESS 2022年第6期30卷 9854-9868页

作者： Wertheimer, Zsolt-Alon Bar, Chen Levin, Anat Technion Israel Inst Technol Dept Elect Engn Haifa Israel

Light propagating through a nonuniform medium scatters as it interacts with particles with different refractive properties such as cells in the tissue. In this work we aim to utilize this scattering process to learn a volumetric reconstruction of scattering parameters, in particular particle densities. We target microscopy applications where coherent speckle effects are an integral part of the imaging process. We argue that the key for successful learning is modeling realistic speckles in the training process. To this end, we build on the development of recent physically accurate speckle simulators. We also explore how to incorporate speckle statistics, such as the memory effect, in the learning framework. Overall, this paper contributes an analysis of multiple aspects of the network design including the learning architecture, the training data and the desired input features. We hope this study will pave the road for future design of learning based imaging systems in this challenging domain. (C) 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement

关键词： image processing Imaging systems Inverse scattering Speckle noise Speckle reduction Three dimensional microscopy

来源：评论

学校读者我要写书评

暂无评论

Evolving Convolutional Neural Networks with Meta-Heuristics for Transfer Learning in Computer vision 3

Evolving Convolutional Neural Networks with Meta-Heuristics ...

引用

3rd International Conference on Evolutionary Computing and Mobile Sustainable Networks, ICECMSN 2023

作者： Srilakshmi, V. Kiran, G. Uday Mounika, M. Sravanthi, A. Sravya, N.V.K. Akhil, V.N.S. Manasa, M. B V Raju Institute of Technology Telangana Narsapur India

In the rapidly evolving landscape of computer vision and artificial intelligence, transfer learning has emerged as a powerful tool for efficiently applying pre-trained models to new tasks. This article delves into the intriguing concept of evolving Convolutional Neural Networks (CNNs) with meta-heuristics for transfer learning in computer vision. The primary focus is on enhancing the adaptability and efficiency of CNNs, making them better suited for specialized tasks. The article covers the significance of transfer learning, the challenges faced in transfer learning with CNNs, the basics of CNN architecture, and the role of meta-heuristics in optimizing CNNs. Real-world applications and success stories demonstrate the transformative potential of these techniques in fields like medical image analysis and autonomous vehicles. It explores emerging trends and potential developments in the domain, emphasizing the impact on various sectors, including healthcare, natural language processing, and robotics. The promise of evolving CNNs with meta-heuristics lies in their capacity to tackle intricate problems with greater precision, ultimately reshaping the landscape of artificial intelligence and machine learning. Ongoing research ensures a promising future for this amalgamation of technologies, promising breakthroughs that will have a lasting impact on the world of computer vision and beyond. © 2023 Elsevier B.V.. All rights reserved.

关键词： Computer vision

来源：评论

学校读者我要写书评

暂无评论

Advanced Medical image processing for Lung Cancer Detection using machine Learning Algorithms 8

Advanced Medical Image Processing for Lung Cancer Detection ...

引用

8th International Conference on Inventive Computation Technologies, ICICT 2025

作者： Karthika, P. Christobel, T. Papitha Department of Computer Science and Applications SRM Institute of Science and Technology Ramapuram Chennai 89 India Department of Computer Science and Applications SRM Institute of Science and Technology Ramapuram Chennai 89 India

ISBN: (纸本)9798331512248

One of the main causes of cancer-related deaths is lung cancer, and increasing survival rates requires early detection. The use of sophisticated machine learning (ML) algorithms to improve the identification of lung cancer from chest X-rays and CT images is investigated in this study. vision Transformers (ViT), Reinforcement Learning (RL), Generative Adversarial Networks (GANs), Meta-Learning, and Ensemble Learning are some of the state-of-the-art methods we use. Compared to conventional CNNs, By capturing long-range dependencies in medical pictures, vision Transformers (ViT) can improve accuracy by up to 85-96% compared to typical CNN models, especially when it comes to feature extraction and categorization. The optimization of diagnostic workflows using Reinforcement Learning (RL) results in a 90% improvement in decision-making efficiency and adaptive learning capabilities, which greatly enhance real-time picture analysis. Training datasets can be enhanced with Generative Adversarial Networks (GANs), which create realistic synthetic images and increase model generalization by 90-97%. This is crucial in situations where data is scarce or unbalanced. For uncommon or underrepresented cancer cases, meta-learning improves classification accuracy by 90% by allowing models to learn from sparsely labeled data. Ensemble learning reduces bias and variation by combining many models using approaches like XGBoost, Bagging, and Stacking, increasing total accuracy by 80-95%. When compared to conventional methods, key performance indicators including precision, recall, and F1 score demonstrate significant gains, with sensitivity rising by up to 90% and specificity improving by 85%. These cutting-edge algorithms greatly improve the detection of lung cancer, facilitating quicker and more precise diagnoses and assisting clinicians in making decisions that will benefit patients. © 2025 IEEE.

关键词： Ensemble Learning Generative Adversarial Networks (GANs) Lung Cancer Detection machine Learning Algorithms Meta-Learning vision Transformers (ViT)

来源：评论

学校读者我要写书评

暂无评论

Outlier-Robust Estimation: Hardness, Minimally Tuned Algorithms, and applications

引用

IEEE TRANSACTIONS ON ROBOTICS 2022年第1期38卷 281-301页

作者： Antonante, Pasquale Tzoumas, Vasileios Yang, Heng Carlone, Luca MIT Lab Informat & Decis Syst Cambridge MA 02139 USA Univ Michigan Dept Aerosp Engn Ann Arbor MI 48109 USA

Nonlinear estimation in robotics and vision is typically plagued with outliers due to wrong data association or incorrect detections from signal processing and machine learning methods. This article introduces two unifying formulations for outlier-robust estimation, generalized maximum consensus ($\textrm{G}$-$\textrm{MC}$) and generalized truncated least squares ($\textrm{G-TLS}$), and investigates fundamental limits, practical algorithms, and applications. Our first contribution is a proof that outlier-robust estimation is inapproximable: In the worst case, it is impossible to (even approximately) find the set of outliers, even with slower-than-polynomial-time algorithms (particularly, algorithms running in quasi-polynomial time). As a second contribution, we review and extend two general-purpose algorithms. The first, adaptive trimming ($\textrm{ADAPT}$), is combinatorial and is suitable for $\textrm{G}$-$\textrm{MC}$;the second, graduated nonconvexity ($\textrm{GNC}$), is based on homotopy methods and is suitable for $\textrm{G-TLS}$. We extend $\textrm{ADAPT}$ and $\textrm{GNC}$ to the case where the user does not have prior knowledge of the inlier-noise statistics (or the statistics may vary over time) and is unable to guess a reasonable threshold to separate inliers from outliers (as the one commonly used in RANdom SAmple Consensus $(\textrm{RANSAC})$. We propose the first minimally tuned algorithms for outlier rejection, which dynamically decide how to separate inliers from outliers. Our third contribution is an evaluation of the proposed algorithms on robot perception problems: mesh registration, image-based object detection (shape alignment), and pose graph optimization. $\textrm{ADAPT}$ and $\textrm{GNC}$ execute in real time, are deterministic, outperform $\textrm{RANSAC}$, and are robust up to 80-90% outliers. Their minimally tuned versions also compare favorably with the state of the art, even though they do not rely on a noise bound for the inliers.

关键词： Estimation Signal processing algorithms Probabilistic logic Approximation algorithms Simultaneous localization and mapping Particle measurements Measurement uncertainty Algorithms autonomous systems computational complexity computer vision maximum likelihood estimation resilient perception robust estimation

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：