The proceedings contain 106 papers. The topics discussed include: macro-AUC-driven active learning strategy for multi-label classification enhancement;mitigating privacy threats without degrading visual quality of VR ...
ISBN:
(纸本)9798350351422
The proceedings contain 106 papers. The topics discussed include: macro-AUC-driven active learning strategy for multi-label classification enhancement;mitigating privacy threats without degrading visual quality of VR applications: using re-identification attack as a case study;attenuation-aware weighted optical flow with medium transmission map for learning-based visual odometry in underwater terrain;GeoVQA: a comprehensive multimodal geometry dataset for secondary education;pulse of the crowd: quantifying crowd energy through audio and video analysis;automated recognition of optic disc and blood vessels in diabetic fundoscopy images using real-timeimage analysis;GeoSecure-B: a method for secure bearing calculation;and exploiting correlation between facial action units for detecting deepfake videos.
Traditional information hiding techniques alter carriers to embed secret information, which steganalysis algorithms can find. In the area of covert communication, coverless information concealment has been suggested a...
详细信息
ISBN:
(纸本)9783031581809;9783031581816
Traditional information hiding techniques alter carriers to embed secret information, which steganalysis algorithms can find. In the area of covert communication, coverless information concealment has been suggested as a way of preventing steganalysis. A new methodology for sharing secret data through coverless video steganography using Local Binary Pattern (LBP) before bit-plane segmentation has been proposed. In this technique, a single frame is converted into multiple bit-planes, and various hash sequences are generated from these bit-planes. After extracting frames from the video, each frame is converted into grayscale using LBP and then split into multiple bit-planes using Bit-plane Complexity Segmentation (BPCS). The hash sequences are obtained by calculating the average median values of corresponding bit-plane sub-blocks. A retrieval database is created to relate the obtained hash sequences with the bit-plane features. And for the first time security analysis is done in this proposed technique. The experimental results demonstrate that this approach achieves better robustness against various attacks, has a larger capacity, requires less time to extract hash sequences, and has a higher success rate of concealing information than existing coverless video steganography techniques.
Federated learning inherently provides a certain level of privacy protection, which however is often inadequate in many real-world scenarios. Existing privacy-preserving methods frequently incur unbearable time overhe...
详细信息
ISBN:
(纸本)9798331529543;9798331529550
Federated learning inherently provides a certain level of privacy protection, which however is often inadequate in many real-world scenarios. Existing privacy-preserving methods frequently incur unbearable time overheads or result in non-negligible deterioration to model performance, thus suffering from the tradeoff between performance and privacy. In this work, we propose a novel Federated Privacy-Preserving Knowledge Transfer framework, namely FedPPKT, which employs data-free knowledge distillation in a meta-learning manner to rapidly generates pseudo data and performs privacy-preserving knowledge transfer. FedPPKT establishes a protective barrier between the original private data and the federated model, thereby ensuring user privacy. Furthermore, leveraging the few-round strategy of FedPPKT, it has the capability to reduce the number of communication rounds, further mitigating the risk of privacy exposure for user data. With the help of the meta generator, the problem of uneven local label distribution on clients is alleviated, mitigating data heterogeneity and improving model performance. Experiments show that FedPPKT outperforms the state-of-the-art privacy-preserving federated learning methods. Our code is publicly available at https://***/HIT-weiqb/FedPPKT.
The proceedings contain 114 papers. The topics discussed include: application of artificial neural networks for processing some biomedical data;distinguishing between AI images and realimages with hybrid image classi...
ISBN:
(纸本)9798350387568
The proceedings contain 114 papers. The topics discussed include: application of artificial neural networks for processing some biomedical data;distinguishing between AI images and realimages with hybrid image classification methods;securing Durres Port's digital transformation: cybersecurity strategy for maritime industry;linguistic encryption for underwater communication;a toolset for blood pressure visualization and measurement in time, frequency and time-frequency domains;using a shape from polarization to determine the 3D surface of objects with thermal radiation;on the influence of cell libraries and other parameters to SCA resistance of crypto IP cores;integration of PXROS-HR with micro-ROS in robotic systems;and traffic-aware video streaming topology reconfiguration for smart city applications.
Fall detection systems use a number of different technologies to achieve their goals. This way, they contribute to better life conditions for the elderly community. The artificial vision is one of these technologies a...
详细信息
Fall detection systems use a number of different technologies to achieve their goals. This way, they contribute to better life conditions for the elderly community. The artificial vision is one of these technologies and, within this field, it has gained momentum over the course of the last few years as a consequence of the incorporation of different artificial neural networks (ANN's). These ANN's share a common characteristic, they are used to extract descriptors from images and video clips that, properly processed, will determine whether a fall has taken *** descriptors, which capture kinematic features associated with the fall, are inferred from datasets recorded by young volunteers or actors who simulate falls. Systems based on this concept offer excellent performances in tests which use that kind of datasets. However, given the well-documented differences between these falls and the real ones, concerns about system performances when processing falls of elderly people are *** work implements an alternative approach to the classical use of kinematic descriptors. To do it, for the first time to the best of the authors' knowledge, the authors propose the introduction of human dynamic stability descriptors used in other fields to determine whether a fall has taken place. These descriptors approach the human body in terms of balance and stability;this way, differences between real and simulated falls become irrelevant, as all falls are a direct result of fails in the continuous effort of the body to keep balance, regardless of other considerations. The descriptors are determined by using the information provided by a neural network able to estimate the body centre of mass and the feet projections onto the ground plane, as well as the feet contact *** theory behind this new approach and its validity is studied in this article with very promising results, as it is able to match or over exceed the performances of previous systems using kinematic de
The aim of this research is to explore methods for building architectural virtual scenes based on imageprocessing techniques to meet the needs of the fields of architectural design, visualization and simulation. With...
详细信息
A 1 millisecond (1 -ms) vision system, which processes videos at 1000 frames per second (FPS) within 1 ms/frame delay, plays an increasingly important role in fields such as robotics and factory automation. Superpixel...
详细信息
A 1 millisecond (1 -ms) vision system, which processes videos at 1000 frames per second (FPS) within 1 ms/frame delay, plays an increasingly important role in fields such as robotics and factory automation. Superpixel as one of the most extensively employed image oversegmentation methods is a crucial pre-processing step for reducing computations in various computer vision applications. Among the different superpixel methods, simple linear iterative clustering (SLIC) has gained widespread adoption due to its simplicity, effectiveness, and computational efficiency. However, the iterative assignment and update steps in SLIC make it challenging to achieve high processing speed. To address this limitation and develop a SLIC superpixel segmentation system with a 1 ms delay, this paper proposes grid sample based temporal iteration. By leveraging the high frame rate of the input video, the proposed method distributes the iterations into the temporal domain, ensuring that the system's delay keeps within one frame. Additionally, grid sample information is added as initialization information to the obtained superpixel centers for enhancing the stability of superpixels. Furthermore, a selective label propagation based pipeline architecture is proposed for parallel computation of all the possibilities of label propagation. This eliminates data dependency between adjacent pixels and enables a fully pipelined system. The evaluation results demonstrate that the proposed superpixel segmentation system achieves boundary recall and under -segmentation error comparable to the original SLIC algorithm. When considering label consistency, the proposed system surpasses the performance of state-of-the-art superpixel segmentation methods. Moreover, in terms of hardware performance, the proposed system processes 1000 FPS images with 0.985 ms/frame delay.
The agriculture sector is crucial to many economies, particularly in developing regions, with post-harvest technology emerging as a key growth area. The oleaster, valued for its nutritional and medicinal properties, h...
详细信息
The agriculture sector is crucial to many economies, particularly in developing regions, with post-harvest technology emerging as a key growth area. The oleaster, valued for its nutritional and medicinal properties, has traditionally been graded manually based on color and appearance. As global demand rises, there is a growing need for efficient automated grading methods. Therefore, this study aimed to develop a real-time machine vision system for classifying oleaster fruit at various grading velocities. Initially, in the offline phase, a dataset containing video frames of four different quality classes of oleaster, categorized based on the Iranian national standard, was acquired at different linear conveyor belt velocities (ranging from 4.82 to 21.51 cm/s). The Mask R-CNN algorithm was used to segment the extracted frames to obtain the position and boundary of the samples. Experimental results indicated that, with a 100% detection rate and an average instance segmentation accuracy error ranging from 4.17 to 5.79%, the Mask R-CNN algorithm is capable of accurately segmenting all classes of oleaster at all the examined grading velocity levels. The results of the fivefold cross validation indicated that the general YOLOv8x and YOLOv8n models, created using the dataset obtained from all conveyor belt velocity levels, have a similarly reliable classification performance. Therefore, given its simpler architecture and lower processingtime requirements, the YOLOv8n model was used to evaluate the grading system in real-time mode. The overall classification accuracy of this model was 92%, with a sensitivity range of 87.10-94.89% for distinguishing different classes of oleaster at a grading velocity of 21.51 cm/s. The results of this study demonstrate the effectiveness of deep learning-based models in developing grading machines for the oleaster fruit.
In the past few years, several efforts have been devoted to reduce individual sources of latency in video delivery, including acquisition, coding and network transmission. The goal is to improve the quality of experie...
详细信息
ISBN:
(数字)9781665496209
ISBN:
(纸本)9781665496209
In the past few years, several efforts have been devoted to reduce individual sources of latency in video delivery, including acquisition, coding and network transmission. The goal is to improve the quality of experience in applications requiring real-time interaction. Nevertheless, these efforts are fundamentally constrained by technological and physical limits. In this paper, we investigate a radically different approach that can arbitrarily reduce the overall latency by means of video extrapolation. We propose two latency compensation schemes where video extrapolation is performed either at the encoder or at the decoder side. Since a loss of fidelity is the price to pay for compensating latency arbitrarily, we study the latency-fidelity compromise using three recent video prediction schemes. Our preliminary results show that by accepting a quality loss, we can compensate a typical latency of 100 ms with a loss of 8 dB in PSNR with the best extrapolator. This approach is promising but also suggests that further work should be done in video prediction to pursue zero-latency video transmission.
During harsh weather conditions, the presence of fog, dust in the environment degrades the image's quality, which affects the visibility of drivers of heavy earth-moving machinery in opencast mines. Due to low vis...
详细信息
During harsh weather conditions, the presence of fog, dust in the environment degrades the image's quality, which affects the visibility of drivers of heavy earth-moving machinery in opencast mines. Due to low visibility, mining operations cannot be carried out as drivers are easily prone to accidents. This paper proposes a technique that includes developing a vision enhancement system called perceptive driving assistant system for increasing visibility of real-timevideo of the road in front of the vehicle for operators of heavy earth-moving machinery at opencast mines during harsh weather conditions to overcome the problem. The system consists of high-quality Internet Protocol cameras and thermal cameras for real-timeimageprocessing and other well-defined devices, which is quite capable of enhancing the visibility of the image, outlining edges of the road, and detecting obstacles present on the path of operators for smooth driving and reducing threat of accidents. A high-speed graphical processing unit has been used for quality-performance parallel computing, which is well suited for real-time operations to empower fast real-time operations. The calculated frame per second (fps) of image enhancement, object detection, and edge detection is 17.91, 15.91, and 25.09 fps, respectively. The actual frame rate is 26.07 fps, and after applying the algorithm, the final frame rate is 19.65 fps. The calculated accuracy of the object detection model is 81.23%. Field trials indicate that the developed system has performed adequately during foggy weather.
暂无评论