Video based facial expression recognition has been a long standing problem and attracted growing attention recently. The key to successful system is exploit the potentials of audiovisual modalities design robust features effectively characterize appearance configuration changes caused by motions. We propose an effective framework address this issue in paper. In our study, both visual (face images) audio (speech) are utilized. A new feature descriptor called Histogram Oriented Gradients from...

The authors present an efficient two-stage approach for leaf image retrieval by using simple shape features including centroid–contour distance (CCD) curve, eccentricity and angle code histogram (ACH). In the first stage, images that are dissimilar with query will be filtered out to reduce search space, fine follow all three sets of in reduced space second stage. Different from ACH, CCD curve is neither scaling-invariant nor rotation-invariant. Therefore, normalisation required achieve...

Sleep staging is to score the sleep state of a subject into different stages such as Wake and Rapid Eye Movement (REM). It plays an indispensable role in diagnosis treatment disorders. As manual through well-trained experts time consuming, tedious, subjective, many automatic methods have been developed for accurate, efficient, objective staging. Recently, deep learning based successfully proposed electroencephalogram (EEG) with promising results. However, most these directly take EEG raw...

This paper presents our proposed approach for the second Emotion Recognition in The Wild Challenge. We propose a new feature descriptor called Histogram of Oriented Gradients from Three Orthogonal Planes (HOG_TOP) to represent facial expressions. also explore properties visual features and audio features, adopt Multiple Kernel Learning (MKL) find an optimal fusion. An SVM with multiple kernels is trained expression classification. Experimental results demonstrate that method achieves...

Although promising results have been achieved for human action recognition under well-controlled conditions, it is very challenging to recognize actions in realistic scenarios due increased difficulties such as dynamic backgrounds. In this paper, we propose take multimodal (i.e., audiovisual) characteristics of videos into account the first time, since, scenarios, audio signals accompanying an generally provide a cue nature action, phone ringing answering . order cope with diverse cues...

Strabismus is one of the most common vision diseases that would cause amblyopia and even permanent loss. Timely diagnosis crucial for well treating strabismus. In contrast to manual diagnosis, automatic recognition can significantly reduce labor cost increase efficiency. this paper, we propose recognize strabismus using eye-tracking data convolutional neural networks. particular, an eye tracker first exploited record a subject’s movements. A gaze deviation (GaDe) image then proposed...

This letter proposes a novel neural root finder based on the moment method (RMM) to find arbitrary roots (including complex ones) of polynomials. (NRF) was designed feedforward networks (FNN) and trained with constrained learning algorithm (CLA). Specifically, we have incorporated priori information about moments polynomials into conventional backpropagation (BPA), construct new CLA. The resulting NRF is shown be able rapidly estimate distributions We study compare advantage RMM-based over...

This paper proposes new modified constrained learning neural root finders (NRFs) of polynomial constructed by backpropagation network (BPN). The technique is based on the relationships between roots and coefficients as well moments polynomial. We investigated different resulting algorithms (CLAs) variants error cost functions (ECFs) in BPN derived a CLA (MCLA), found that computational complexities MCLA root-moment method (RMM) are order polynomial, simpler than CLA. Further, we also...

Living plant recognition based on images of leaf, flower and fruit is a very challenging task in the field pattern computer vision. There has been little work reported image processing recognition. In recent years, several researchers have dedicated their to leaf characterisation. As an inherent trait, vein definitely contains important information for species despite its complex modality. A new approach that combines thresholding method artificial neural network (ANN) classifier proposed...

Strabismus is one of the most common vision disorders in preschool children. It can cause amblyopia and even permanent loss. In addition to a problem, strabismus brings both children adults serious negative impacts their daily life, education, employment etc. Timely diagnosis thus crucial. However, traditional methods conducted by ophthalmologists rely significantly on experiences, making results subjective. also inconvenient for those being used examination large communities such as...

Document Image Recognition (DIR), a very useful technique in office automation and digital library applications, is to find the most similar template for any input document image prestored data set. Existing methods use both local features global layout information. In this paper, we propose novel algorithm based on matching of Component Block Projections (CBP), which are concatenated directional projection vectors component blocks image. Compared those existing methods, CBP-based...

The purpose of this work is to develop an interactive tool which helps botanists extract the vein system with its hierarchical properties as little user interaction possible. In paper, we present a new venation extraction method using independent component analysis (ICA). popular and efficient FastICA algorithm applied patches leaf images learn set linear basis functions or features for then are used pattern map extraction. our experiments, training sets randomly generated from different...

Recently, biometric identification techniques have attracted great attention due to increasing demand of high-performance security systems. Compared with conventional methods, provide more reliable and robust solutions. In this paper, a novel video-based model based on eye tracking technique is proposed. Inspired by visual attention, video clips are designed for subjects view in order capture data reflecting their physiological behavioral characteristics. Various characteristics, including...

The timely and accurate identification of plant species is a persistent challenge as pressure from human activity threatens global flora biodiversity. Most existing research on computer based has focused using leaf contour, signature spectral analysis techniques alongside textural properties the lamina. However, these feature methods often suffer limited discriminative ability scalability, particularly mobile devices for use in field. In this paper, we propose novel descriptor named EAGLE...