A recurring problem faced when training neural networks is that there typically not enough data to maximize the generalization capability of deep networks. There are many techniques address this, including augmentation, dropout, and transfer learning. In this paper, we introduce an additional method, which call smart augmentation show how use it increase accuracy reduce over fitting on a target network. Smart works, by creating network learns generate augmented during process in way reduces...

In the last few years, we have witnessed an exponential growth in research activity into advanced training of convolutional neural networks (CNNs), a field that has become known as deep learning. This been triggered by combination availability significantly larger data sets, thanks part to corresponding big data, and arrival new graphics-processing-unit (GPU)-based hardware enables these large sets be processed reasonable timescales. Suddenly, wide variety long-standing problems machine...

Current activities in the Internet of Things (IoT) are focused on architectures, protocols, and networking for efficient interconnection heterogeneous things, infrastructure deployment, creation value-added services. The majority IoT products, services, platforms supported by cloud-computing platforms. With being a multidisciplinary ecosystem, it is now utilized connection with scenarios demanding real-time data processing feedback, example, connected autonomous vehicles scenarios. Cloud not...

Real time face detection combined with eyegaze tracking can provide a means of user input into gaming environment. Game and CE system designers use facial eye-gaze information in various ways to enhance UI design providing smarter modes gameplay interaction modalities that are sensitive users behaviors mood. Here we review earlier approaches, using wearable sensors, or enhanced IR illumination. Our technique only requires video feed from low-resolution user-facing camera. The algorithm is...

Blockchain technology is ready to disrupt nearly every industry and business model, the energy sector no exception. Energy businesses across world have already started exploring use of blockchain in large-scale trading systems, peer-to-peer trading, project financing, supply chain tracking, asset management among other applications. Information Communication Technologies (ICTs) recently revolutionizing landscape, now providing an additional opportunity make system more intelligent,...

Taking inspiration from recent developments in visual generative tasks using diffusion models, we propose a method for end-to-end speech-driven video editing denoising model. Given of talking person, and separate auditory speech recording, the lip jaw motions are re-synchronised without relying on intermediate structural representations such as facial landmarks or 3D face We show this is possible by conditioning model audio mel spectral features to generate synchronised motion. Proof concept...

A statistical active appearance model (AAM) is developed to track and detect eye blinking. The has been designed be robust variations of head pose or gaze. In particular we analyze determine the parameters which encode caused by This global further extended using a series sub-models enable independent modeling tracking two regions. Several methods measurement detection eye-blink are proposed evaluated. results various tests on different image databases presented validate each model.

Accurate and efficient eye gaze estimation is important for emerging consumer electronic systems, such as driver monitoring systems novel user interfaces. Such are required to operate reliably in difficult, unconstrained environments with low power consumption at minimal cost. In this paper, a new hardware friendly, convolutional neural network (CNN) model computational requirements introduced assessed appearance-based estimation. The tested compared against existing CNN approaches,...

AI-based smart thermal perception systems can cater to the limitations of conventional imaging sensors by providing a more reliable data source in low-lighting conditions and adverse weather conditions. This research evaluates modifies state-of-the-art object detection classifier framework for vision with seven key classes order provide superior sensing scene understanding input advanced driver-assistance (ADAS). The networks are trained on public datasets is validated test three different...

Event cameras, unlike traditional frame-based excel in detecting and reporting changes light intensity on a per-pixel basis. This unique technology offers numerous advantages, including high temporal resolution, low latency, wide dynamic range, reduced power consumption. These characteristics make event cameras particularly well-suited for sensing applications such as monitoring drivers or human behavior. paper presents feasibility study the using multitask neural network with real-time...

The explosion of user-generated content on the Social Web published from mobile devices has seen reporting applications grow in popularity self-reporting style workout tracking like Fitocracy <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> to community FixMyStreet xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> . In this paper, we introduce concept social where a people report issues within their environment. We discuss annotating...

As a near ideal biometric, iris authentication is widely used and mobile acquisition techniques are known. But on handheld imaging devices, such as smartphones, poses multiple, unique challenges. In this paper, range of factors that affect the quality images reviewed. Iris size, image wavelength found to be key factors. Experimental results presented confirming lower limits size for useful performance. The workflow devices described. A case study current smartphone model presented, including...

The technology to connect ?things? the Internet has existed for more than 20 years, so if we take a look back at recent history, might well be tempted ask question ?Why now?? In this article, I address that question, examine origins of Things (IoT), and forward next wave disruptive technologies will coming device near you in few years.

With the advent of deepfake videos, video forgery has become a serious threat. Videos in social media are most common and targets. There some existing works for detecting videos but very few attempts have been made media. This paper presents neural network based method to detect fake videos. A model, consisting convolutional (CNN) classifier is proposed. Three different structures, XceptionNet, InceptionV3 Resnet50 considered as CNN modules comparative study made. Xception Net chosen...

Despite recent advancements in deep learning technologies, Child Speech Recognition remains a challenging task. Current Automatic (ASR) models require substantial amounts of annotated data for training, which is scarce. In this work, we explore using the ASR model, wav2vec2, with different pretraining and finetuning configurations self-supervised (SSL) toward improving automatic child speech recognition. The pretrained wav2vec2 were finetuned training data, adult combination both, to...

A user centric approach to authentication for home networks is proposed. zero-knowledge-proof (ZKP) used leverage the emerging cloud infrastructure allowing users temporarily transfer their service and content rights within a trusted environment such as friend's home. This enables sharing of personalized more sophisticated network-based services over conventional TCP/IP infrastructure. Experimental results derived from reference prototype are presented. These demonstrate practicality...

Structural obfuscation offers a means to effectively secure through the contents of an intellectual property (IP) cores used in electronic system-on-chip (SoC). In this work novel structural methodology for protecting digital signal processor (DSP) IP core at architectural synthesis design stage. The proposed approach specifically targets protection that involve complex loops. Five different algorithmic level transformation techniques are employed: loop unrolling, invariant code motion, tree...

By growing the capacity and processing power of handheld devices nowadays, a wide range capabilities can be implemented in these to make them more intelligent user friendly. Determining mood used order provide suitable reactions from device different conditions. One most studied ways detection is by using facial expressions, which still one challenging fields pattern recognition machine learning science. Deep Neural Networks (DNN) have been widely overcome difficulties expression...