Optimization and Performance Analysis of Intelligent Video AI Dynamic

In today’s information age,video data,as an important carrier of information,is growing explosively in terms of production volume.The quick and accurate extraction of useful information from massive video data has become a focus of research in the field of computer vision.AI dynamic recognition technology has become one of the key technologies to address this issue due to its powerful data processing capabilities and intelligent recognition functions.Based on this,this paper first elaborates on the development of intelligent video AI dynamic recognition technology,then proposes several optimization strategies for intelligent video AI dynamic recognition technology,and finally analyzes the performance of intelligent video AI dynamic recognition technology for reference.

CVTD: A Robust Car-Mounted Video Text Detector

Text perception is crucial for understanding the semantics of outdoor scenes,making it a key requirement for building intelligent systems for driver assistance or autonomous driving.Text information in car-mounted videos can assist drivers in making decisions.However,Car-mounted video text images pose challenges such as complex backgrounds,small fonts,and the need for real-time detection.We proposed a robust Car-mounted Video Text Detector(CVTD).It is a lightweight text detection model based on ResNet18 for feature extraction,capable of detecting text in arbitrary shapes.Our model efficiently extracted global text positions through the Coordinate Attention Threshold Activation(CATA)and enhanced the representation capability through stacking two Feature Pyramid Enhancement Fusion Modules(FPEFM),strengthening feature representation,and integrating text local features and global position information,reinforcing the representation capability of the CVTD model.The enhanced feature maps,when acted upon by Text Activation Maps(TAM),effectively distinguished text foreground from non-text regions.Additionally,we collected and annotated a dataset containing 2200 images of Car-mounted Video Text(CVT)under various road conditions for training and evaluating our model’s performance.We further tested our model on four other challenging public natural scene text detection benchmark datasets,demonstrating its strong generalization ability and real-time detection speed.This model holds potential for practical applications in real-world scenarios.

Video Recommendation System Using Machine-Learning Techniques

In the realm of contemporary artificial intelligence,machine learning enables automation,allowing systems to naturally acquire and enhance their capabilities through learning.In this cycle,Video recommendation is finished by utilizing machine learning strategies.A suggestion framework is an interaction of data sifting framework,which is utilized to foresee the“rating”or“inclination”given by the different clients.The expectation depends on past evaluations,history,interest,IMDB rating,and so on.This can be carried out by utilizing collective and substance-based separating approaches which utilize the data given by the different clients,examine them,and afterward suggest the video that suits the client at that specific time.The required datasets for the video are taken from Grouplens.This recommender framework is executed by utilizing Python Programming Language.For building this video recommender framework,two calculations are utilized,for example,K-implies Clustering and KNN grouping.K-implies is one of the unaided AI calculations and the fundamental goal is to bunch comparable sort of information focuses together and discover the examples.For that K-implies searches for a steady‘k'of bunches in a dataset.A group is an assortment of information focuses collected due to specific similitudes.K-Nearest Neighbor is an administered learning calculation utilized for characterization,with the given information;KNN can group new information by examination of the‘k'number of the closest information focuses.The last qualities acquired are through bunching qualities and root mean squared mistake,by using this algorithm we can recommend videos more appropriately based on user previous records and ratings.

Guest Editorial:Intelligent Video Surveillance and Related Technologies

Due to the increasing demand for developing a secure and smart living environment, the intelligent video surveillance technology has attracted considerable attention. Building an automatic, reliable, secure, and intelligent video surveillance system has spawned large research projects and triggered many popular research topics in several international conferences and workshops recently. This special issue of Journal of ElecWonic Science and Technology （JEST） aims to present recent advances in video surveillance systems which address the observation of people in an environment, leading to a real-time description of their actions and interactions.

Semi-automatic Video Annotation Tool to Generate Ground Truth for Intelligent Video Surveillance Systems

Generating ground truth data for developing object detection algorithms of intelligent surveillance systems is a considerably important yet time-consuming task; therefore, a user-friendly tool to annotate videos efficiently and accurately is required. In this paper, the development of a semi-automatic video annotation tool is described. For efficiency, the developed tool can automatically generate the initial annotation data for the input videos utilizing automatic object detection modules, which are developed independently and registered in the tool. To guarantee the accuracy of the ground truth data, the system also has several user-friendly functions to help users check and edit the initial annotation data generated by the automatic object detection modules. According to the experiment＇s results, employing the developed annotation tool is considerably beneficial for reducing annotation time; when compared to manual annotation schemes, using the tool resulted in an annotation time reduction of up to 2.3 times.

智能IV诊断在分析光伏低效子阵中的应用

由于光伏子阵的非线性行为和其对运行环境的依赖性,传统的保护装置可能无法诊断光伏系统低效子阵问题。为此,业内进行了大量研究来克服这一问题的出现。然而,大多数方法不仅耗时较久,还存在较大的电量损失。同时,由于需要大量数据集,还存在数据过度拟合问题,导致准确率有所欠缺。在本研究中,通过应用华为AI BOOST智能IV诊断3.0,无论是实验评估还是实际应用,相比传统的诊断方法,均取得了较为明显的效果。

Research on Big Data and Artificial Intelligence Aided Decision-Making Mechanism with the Applications on Video Website Homemade Program Innovation

In this paper, we conduct research on the big data and the artificial intelligence aided decision-making mechanism with the applications on video website homemade program innovation. Make homemade video shows new media platform site content production with new possible, as also make the traditional media found in Internet age, the breakthrough point of the times. Site homemade video program, which is beneficial to reduce copyright purchase demand, reduce the cost, avoid the homogeneity competition, rich advertising marketing at the same time, improve the profit pattern, the organic combination of content production and operation, complete the strategic transformation. On the basis of these advantages, once the site of homemade video program to form a brand and a higher brand influence. Our later research provides the literature survey for the related issues.

System Structure and Calibration Models of Intelligent Photogrammetron

This paper describes the structure, geometric model and geometric calibrationof Photogrammetron Ⅰ - the first type of photogrammetron which is designed to be a coherent stereophotogrammetric system in which two cameras are mounted on a physical base but driven by anintelligent agent architecture. The system calibration is divided into two parts: the in-labcalibration determines the fixed parameters in advance of system operation, and the in-situcalibration keeps tracking the free parameters in real-time during the system operation. In a videosurveillance set-up, prepared control points are tracked in stereo image sequences, so that the freeparameters of the system can be continuously updated through iterative bundle adjustment and Kalmanfiltering.

High-Movement Human Segmentation in Video Using Adaptive N-Frames Ensemble

Awide range of camera apps and online video conferencing services support the feature of changing the background in real-time for aesthetic,privacy,and security reasons.Numerous studies show that theDeep-Learning(DL)is a suitable option for human segmentation,and the ensemble of multiple DL-based segmentation models can improve the segmentation result.However,these approaches are not as effective when directly applied to the image segmentation in a video.This paper proposes an Adaptive N-Frames Ensemble(AFE)approach for high-movement human segmentation in a video using an ensemble of multiple DL models.In contrast to an ensemble,which executes multiple DL models simultaneously for every single video frame,the proposed AFE approach executes only a single DL model upon a current video frame.It combines the segmentation outputs of previous frames for the final segmentation output when the frame difference is less than a particular threshold.Our method employs the idea of the N-Frames Ensemble(NFE)method,which uses the ensemble of the image segmentation of a current video frame and previous video frames.However,NFE is not suitable for the segmentation of fast-moving objects in a video nor a video with low frame rates.The proposed AFE approach addresses the limitations of the NFE method.Our experiment uses three human segmentation models,namely Fully Convolutional Network(FCN),DeepLabv3,and Mediapipe.We evaluated our approach using 1711 videos of the TikTok50f dataset with a single-person view.The TikTok50f dataset is a reconstructed version of the publicly available TikTok dataset by cropping,resizing and dividing it into videos having 50 frames each.This paper compares the proposed AFE with single models and the Two-Models Ensemble,as well as the NFE models.The experiment results show that the proposed AFE is suitable for low-movement as well as high-movement human segmentation in a video.

Artificial intelligence as a means to improve recognition of gastrointestinal angiodysplasia in video capsule endoscopy

Gastrointestinal angiodysplasia(GIAD)is defined as the pathological process where blood vessels,typically venules and capillaries,become engorged,tortuous and thin walled–which then form arteriovenous connections within the mucosal and submucosal layers of the gastrointestinal tract.GIADs are a significant cause of gastrointestinal bleeding and are the main cause for suspected small bowel bleeding.To make the diagnosis,gastroenterologists rely on the use of video capsule endoscopy(VCE)to“target”GIAD.However,the use of VCE can be cumbersome secondary to reader fatigue,suboptimal preparation,and difficulty in distinguishing images.The human eye is imperfect.The same capsule study read by two different readers are noted to have miss rates like other forms of endoscopy.Artificial intelligence(AI)has been a means to bridge the gap between human imperfection and recognition of GIAD.The use of AI in VCE have shown that detection has improved,however the other burdens and limitations still need to be addressed.The use of AI for the diagnosis of GIAD shows promise and the changes needed to enhance the current practice of VCE are near.

Deepfake Video Detection Employing Human Facial Features

Deepfake technology can be used to replace people’s faces in videos or pictures to show them saying or doing things they never said or did. Deepfake media are often used to extort, defame, and manipulate public opinion. However, despite deepfake technology’s risks, current deepfake detection methods lack generalization and are inconsistent when applied to unknown videos, i.e., videos on which they have not been trained. The purpose of this study is to develop a generalizable deepfake detection model by training convoluted neural networks (CNNs) to classify human facial features in videos. The study formulated the research questions: “How effectively does the developed model provide reliable generalizations?” A CNN model was trained to distinguish between real and fake videos using the facial features of human subjects in videos. The model was trained, validated, and tested using the FaceForensiq++ dataset, which contains more than 500,000 frames and subsets of the DFDC dataset, totaling more than 22,000 videos. The study demonstrated high generalizability, as the accuracy of the unknown dataset was only marginally (about 1%) lower than that of the known dataset. The findings of this study indicate that detection systems can be more generalizable, lighter, and faster by focusing on just a small region (the human face) of an entire video.

改进YOLOv5s的采煤机滚筒与支架护帮板干涉状态智能识别

针对综采工作面液压支架护帮板处于未收回异常状态导致采煤机滚筒与护帮板干涉问题,提出一种改进YOLOv5s的采煤机滚筒与液压支架护帮板干涉状态智能识别方法。运用课题组前期提出的基于边界约束和非线性上下文正则化的去雾去尘方法对视频图像进行清晰化处理,提高综采工作面监控视频图像质量;对YOLOv5s模型进行改进,通过将YOLOv5s主干网络中的普通卷积Conv替换为分类效果更佳的Ghost卷积,减少了模型的参数数量,提高了模型识别速度,同时引入坐标注意力机制,提高了模型对护帮板和滚筒特征提取能力,从而提高模型识别精确率。运用软非极大值抑制算法(Soft-NMS)的锚框筛选方法,减少因护帮板重叠而发生漏检问题。针对采煤机滚筒与液压支架护帮板干涉状态判定问题,提出液压支架护帮板与采煤机滚筒锚框重合度的判定方法。运用本文改进YOLOv5s模型与YOLOv5s、YOLOv3-tiny模型进行对比分析,结果表明:本文方法与原模型相比的识别精确率提高了约8.1%,GFLOPs降低1.86倍;mAP@.5达到97.2%、平均识别速度为检测时间为5.9 ms。运用本文方法对煤矿实际综采工作面采煤机滚筒与液压支架护帮板视频图像进行干涉状态识别试验验证,结果表明:对采煤机滚筒与液压支架护帮板干涉状态识别准确率为96%。

算法欣赏VS算法厌恶:短视频智能推荐下的用户“算法悖论”

[研究目的]探寻用户在使用人工智能推荐服务时算法厌恶态度与持续使用行为相悖的现象,有助于企业理解用户的算法厌恶效应,优化短视频平台的人工智能推荐服务。[研究方法]基于社会交换理论和技术接受理论,从算法欣赏和算法厌恶的角度构建算法悖论理论模型,采用结构方程模型(SEM)和模糊集定性比较分析(fsQCA)对研究模型进行实证分析。[研究结论]结果发现,短视频平台智能服务中存在用户算法厌恶态度与持续使用行为相悖的“算法悖论”现象。用户通过权衡感知利益和风险来决定持续使用行为,且感知利益可以通过算法欣赏积极影响持续使用行为,而算法厌恶对持续使用行为的影响作用微弱且不显著;平台依赖负向调节算法厌恶与持续使用行为之间的负相关关系,乐观偏差的调节作用不显著;fsQCA分析进一步发现,算法悖论的产生路径可以分为感知利益与风险权衡分析和理性与非理性双重因素驱动两种模式。研究结果为短视频平台中算法悖论现象产生机制提供了理论解释,为优化人工智能推荐在短视频平台的应用提供借鉴。

Sora文生影像模式下中国风格产品系统参数化建构策略研究

Sora文生影像模式的出现,标志着人工智能正式进入千亿级参数量的大数据、大模型时代。AI智能既初步具备了机器知觉能够“识万物”,又能通过输入文字、参数使“万物生”。伴随新质生产力的壮大,中国应深刻理解科技不仅是生产力,而且是意识形态,应努力建构符合中国美学风格、满足中国用户需求的智能化系统;推动中国语境下的科学与人文相结合,形成能够在未来国际虚拟社区传播、具有中国风格的自创生AI体系,全面推进新时代中国智造与世界智能新业态的紧密融合。

面向智能视频监控的空中交通管制员图像分割

为解决复杂场景下空中交通管制员检测与分割精度低、鲁棒性差的问题,提出一种基于掩码区域卷积神经网络(Mask Region-based Convolutional Neural Networks, Mask R-CNN)的管制员图像分割模型ATC Mask R-CNN(ATC Mask Region-based Convolutional Neural Networks)。首先,构建管制员监控图像数据集(ATC Monitor Image Dataset, AMID)并用于模型训练、测试;其次,在主干网络中引入瓶颈注意力模块(Bottleneck Attention Module, BAM)以增强管制员特征提取,采取改进的柔性非极大值抑制算法(Soft Non-maximum Suppression, Soft-NMS)替代NMS算法进行候选框选取,提高对遮挡目标的检测分割;最后,基于AMID进行管制员图像分割试验。结果显示:ATC Mask R-CNN的精确率、召回率和平均精度分别为96.49%、95.62%和88.84%,表明了该方法的有效性。与Mask R-CNN相比,ATC Mask R-CNN有效降低了复杂场景的不利影响,更适用于管制员工作场景,可以为管制大厅安全管理自动化应用提供技术支撑。

视频侦查技术关键及其发展展望

视频侦查技术的核心是通过视频图像的提取、查看、分析和研判来获取侦查线索、固定涉案证据。多年来经过在实战应用中不断地发展创新,视频侦查技术形成了自己独特的关键技术体系:视频信息分析解读技术是侦查应用和证据固定的基础和核心;视频证据固定保全技术的规范是审判中心主义的客观要求,可以获取视频侦查记录报告、视频检验鉴定报告或视频数据关联报告;低质量视频图像的增强恢复技术专业性强,应用范围窄,技术成熟度高,然而它不断面临新的挑战。目前,视频数据的智能应用在大数据背景下变得越来越重要,仍需进一步突破视频自动识别技术的应用范畴,建立完善多层次的视频数据综合应用体系,打造适应不同业务需要的视频数据实战应用模型。

人工智能技术驱动视觉传达作品生成研究综述

目的为改进人工智能技术驱动视觉传达作品的生成方式,提升视觉传达作品的生成质量并为视觉传达设计效率提供理论支撑。方法基于Scopus和中国知网数据库下载并整理相关文献,分析现有人工智能技术驱动视觉传达作品生成的关键技术、研究方向,以及研究方法。结论通过精读文献划分出目前人工智能技术驱动视觉传达作品生成的研究方向,包含以文字生成图像、以图像生成图像,以及视频生成。提取各研究方向中所采用的研究方法,涵盖生成对抗网络、知识推理、空间自适应等。通过分析人工智能技术驱动视觉作品生成的研究现状及方向,进一步总结和归纳研究方向和方法,为未来设计师应对复杂设计挑战开辟了新路径,同时为未来人工智能技术赋能视觉传达作品生成提供了参考和依据。

文生视频类人工智能的风险与三维规制:以Sora为视角

文生视频类人工智能Sora一经发布即引发万众瞩目,其具有的强理解能力、高度仿真性及多模态融合能力为社会带来视觉、听觉震撼的同时,引发诸多法律风险。与此前的生成式人工智能大模型相比,Sora的潜在法律风险在人格权保护、网络犯罪及社会信任等方面更为突出。面对前沿科技给个人权益、刑事犯罪及社会稳定等领域带来的挑战,要及时采取相应的多维规制对策。一是加强对人格权的民法保护,明确个人信息使用的授权,强化数据采集和视频内容监管;二是优化刑法适用与归责,完善刑事法律的解释、适用及责任制度;三是通过规范监管,提升社会信任,推动人工智能由规制对象转向规制工具,助推人工智能系统的安全性和可靠性。

“世界模拟”的拟像迷思——基于通用视觉大模型技术的哲学反思

随着通用视觉大模型技术的迅速发展,对人工智能技术底层逻辑的哲学反思变得刻不容缓。生成式人工智能文生视频、文生图像现象的背后,是数字技术借由算法公理化逻辑所营造出的普世视觉景观,这种视觉景观消解了真实与虚拟之间的边界,在本质上与一种通过数字技术而布展的知识生产权力具有同构性。基于数据预训练和投喂的通用视觉大模型,其知识生产中存在着数据“通用”性与模型“泛化”的张力,大模型泛化能力的提高意味着其所依赖的数据来源愈加具有普遍性和公理性,由此在技术无意识层面形成一种代表数字普遍理性的公共知识体系。为此应重新思考虚拟与现实之间的边界问题,在技术设计关注差异要素的基础上,探索人工智能时代人机交互的可能前景。

基于FFmpeg多线程编码的智能交通监控系统设计

针对智能交通管理设备本身缺乏安全监管,传统视频监控延迟高、画质低、稳定性差的问题,提出一种基于FFmpeg的多线程编码视频流传输方案。通过FFmpeg调用h264_nvenc编码器,实现宏块行级的GPU多线程加速,降低编码延迟。使用Visual Studio 2019和QT15.5开发基于FFmpeg的音视频处理软件,对多路视频流进行封装、推流,并搭建Nginx流媒体服务器进行分发。通过实验表明,该系统整体的传输延迟低于1 s,且拥有良好的率失真特性,监控画面清晰、稳定性高,实现了对交通管理设备实时稳定的安全监控。