There has been a growing interest in the use of data-driven regularizers to solve inverse problems associated with computational imaging systems. The convolutional sparse representation model has recently gained atten...
详细信息
ISBN:
(纸本)9781479981311
There has been a growing interest in the use of data-driven regularizers to solve inverse problems associated with computational imaging systems. The convolutional sparse representation model has recently gained attention, driven by the development of fast algorithms for solving the dictionary learning and sparse coding problems for sufficiently large images and data sets. Nevertheless, this model has seen very limited application to tomographic reconstruction problems. In this paper, we present a model-based tomographic reconstruction algorithm using a learnt convolutional dictionary as a regularizer. The key contribution is the use of a data-dependent weighting scheme for the l(1) regularization to construct an effective denoising method that is integrated into the inversion using the Plug-and-Play reconstruction framework. Using simulated data sets we demonstrate that our approach can improve performance over traditional regularizers based on a Markov random field model and a patch-based sparse representation model for sparse and limited-view tomographic data sets.
Traffic density in roads has been increasing day by day which needs intelligent transportation system that can handle the traffic. Traffic management has become inevitable for smart cities. The enormous increase in ve...
详细信息
ISBN:
(数字)9781728151977
ISBN:
(纸本)9781728151984
Traffic density in roads has been increasing day by day which needs intelligent transportation system that can handle the traffic. Traffic management has become inevitable for smart cities. The enormous increase in vehicle numbers has generated more pressure to manage traffic congestion especially during peak hours. If the traffic congestion at a particular point of time can be found, then that information can be useful for managing the traffic in different lanes and change the traffic light cycle dynamically according to the vehicle count in different lanes. In recent years video surveillance and monitoring has been gaining importance. video can be analyzed which can be used to find the traffic density. Many useful information can be obtained by video processing like real time traffic density. vehicle counting can be done by detecting the object, tracking it and then finally counting the objects. Many different techniques are available for object detection and tracking. Deep learning techniques for object detection led to remarkable improvements compared to conventional imageprocessing techniques by removing the weakness in the conventional techniques. This paper provides a survey on various techniques available for vehicle detection and tracking.
Data is the new oil in current technological society. The impact of efficient data has changed benchmarks of performance in terms of speed and accuracy. The enhancement is visualizable because the processing of data i...
详细信息
Data is the new oil in current technological society. The impact of efficient data has changed benchmarks of performance in terms of speed and accuracy. The enhancement is visualizable because the processing of data is performed by two buzzwords in industry called Computer vision (Cv) and Artificial Intelligence (AI). Two technologies have empowered major tasks such as object detection and tracking for traffic vigilance systems. As the features in image increases demand for efficient algorithm to excavate hidden features increases. Convolution Neural Network (CNN) model is designed for urban vehicle dataset for single object detection and YOLOv3 for multiple object detection on KITTI and COCO dataset. Model performance is analyzed, evaluated and tabulated using performance metrics such as True Positive (TP), True Negative (TN), False Positive (FP), False Negative (FN), Accuracy, Precision, confusion matrix and mean Average Precession (mAP). Objects are tracked across the frames using YOLOv3 and Simple Online Real Time Tracking (SORT) on traffic surveillance video. This paper upholds the uniqueness of the state of the art networks like DarkNet. The efficient detection and tracking on urban vehicle dataset is witnessed. The algorithms give real-time, accurate, precise identifications suitable for real-time traffic applications.
Quality criteria are the essence of a measurement system. The goal of assessing the quality of software elements (algorithms) used for imageprocessing is to ensure control of technical performance indicators of the s...
Quality criteria are the essence of a measurement system. The goal of assessing the quality of software elements (algorithms) used for imageprocessing is to ensure control of technical performance indicators of the system as a whole or its individual functional units under the reduction of costs associated with minimizing the loss function (tuning and debugging). The correct choice of individual metrics for the generalized quality indicator to solve tasks in a particular subject area is one of the key steps in system optimization. It ensures the most flexible approach to testing the developed software elements, identifying and eliminating their functional shortcomings. At the same time, we can say that in this way the generalized indicator of the entire system (its goal function) is optimized.
A single training session of tennis requires 30-40 balls and these scattered balls have to be collected at the end of each session which costs time and induces unnecessary physical stress on players. This paper propos...
详细信息
Convolutionary neural networks have been commonly used for object recognition because, by emulating the nervous system’s action in living organisms, they can achieve high precision. Recently, development and deployme...
Convolutionary neural networks have been commonly used for object recognition because, by emulating the nervous system’s action in living organisms, they can achieve high precision. Recently, development and deployment have further boosted the exponential growth of advanced businesses focused on intelligent systems. In specific, different fuel cells have been suggested for deep CNN based on the FPGA platform since it has characteristics of good efficiency, reconfigurability, and rapid growth, respectively. The processor architecture room has not been well explored, although existing FPGA prototypes have shown better performance over standard computers. One crucial concern is that computing efficiency does not fit the memory capacity offered by an FPGA platform. Due to its excellent precision, a Convolutionary computer program sees uses in a range of imageprocessing applications ranging from target identification and tracking to scene comprehension. various algorithms exist for calculating CNNs. Mostly in CPU, three service systems are found: artificial neural, convolutionary artificial neural, genetic analysis, and collective processing. The multiplication uses an out-of-order optimization technique to facilitate simultaneous information flow processing to optimize coarse-grained instruction computing. Next, we suggest the fused structure in CNNs will naturally fuse several layers, repurposing the intermediary details. We explore diverse architectures to optimize the performance of a CNN centered on this working to create. To decide the fusion and algorithm technique for each sheet, we develop an optimum algorithm. We are also designing an integrated toolchain using Xilinx to ease the routing from the template to the FPGA bitstream. Studies using commonly used vGG and Alex Net illustrate that, relative to the previous fusion-based FPGA acceleration for CNNs, our architecture achieves up to 1.5X fine aggregate.
This paper presents FORMap (Fast Ortho Mapping) a simple, automatic, fast and accurate commercial photogrammetry processing software for Unmanned Aerial vehicles (UAv) imagery equiped with Direct Georeferencing (DG) t...
详细信息
Since its creation, the imageNet-1k benchmark set has played a significant role as a benchmark for ascertaining the accuracy of different deep neural net (DNN) models on the image classification problem. Moreover, in ...
详细信息
Since its creation, the imageNet-1k benchmark set has played a significant role as a benchmark for ascertaining the accuracy of different deep neural net (DNN) models on the image classification problem. Moreover, in recent years it has also served as the principal benchmark for assessing different approaches to DNN training. Finishing a 90-epoch imageNet-1k training with ResNet-50 on a NvIDIA M40 GPU takes 14 days. This training requires 10(18) single precision operations in total. On the other hand, the world's current fastest supercomputer can finish 3 x 10(17) single precision operations per second (according to the Nov 2018 Top 500 results). If we can make full use of the computing capability of the fastest supercomputer, we should be able to finish the training in several seconds. Over the last two years, researchers have focused on closing this significant performance gap through scaling DNN training to larger numbers of processors. Most successful approaches to scaling imageNet training have used the synchronous mini-batch stochastic gradient descent (SGD). However, to scale synchronous SGD one must also increase the batch size used in each iteration. Thus, for many researchers, the focus on scaling DNN training has translated into a focus on developing training algorithms that enable increasing the batch size in data-parallel synchronous SGD without losing accuracy over a fixed number of epochs. In this paper, we investigate supercomputers' capability of speeding up DNN training. Our approach is to use a large batch size, powered by the Layer-wise Adaptive Rate Scaling (LARS) algorithm, for efficient usage of massive computing resources. Our approach is generic, as we empirically evaluate the effectiveness on five neural networks: AlexNet, AlexNet-BN, GNMT, ResNet-50, and ResNet-50-v2 trained with large datasets while preserving the state-of-the-art test accuracy. Compared to the baseline of a previous study from Goyal et al. [1] , our approach shows higher
Character recognition is a best case to apply logics from Memetic algorithms (MA) for imageprocessing. In cases, like finger print matching, cent percent accuracy is expected but the character recognition on other ha...
详细信息
Brain signal processing is important for not only the physiologist doing analysis investigation, but also for the clinician inspecting patients, biomedical engineer who is responsible for collecting, processing, and i...
详细信息
Brain signal processing is important for not only the physiologist doing analysis investigation, but also for the clinician inspecting patients, biomedical engineer who is responsible for collecting, processing, and interpreting the electroencephalogram signals by modeling systems and algorithms for their manipulations. The abundant materials on the subject of brain signal/imageprocessing are scattered in different scientific, technological and physiological journals, international conference proceedings, and also in various databases. Therefore, it is altogether a difficult, too time-consuming, and much tiresome work, exclusively to the newcomers in this field. Therefore, this paper focuses on providing the list of popular databases available belonging to the neurological signals, brain signal/image collections, and so on. The count and the kinds of attacks across the networked computer systems have hiked the significance of computer network security. At present, network administrators use to inspect, examine, scrutinize, review, and analyze the network traffic to figure out what is going on and to set up a prompt response in the event of an identified attack. This paper analyzes the different sweep techniques such as Ping sweep, TCP sweep, and Null sweep on the popular databases about the brain signal/image collections. The results of the Ping sweep support status, TCP sweep times, and Null scan times on different servers are discussed finally.
暂无评论