Automatic Face Shape Classification Via Facial Landmark Measurements

This paper tackles the sensitive subject of face shape identification via near neutral-pose 2D images of human subjects. The possibility of extending to 3D facial models is also proposed, and would alleviate the need for the neutral stance. Accurate face shape classification serves as a vital building block of any hairstyle and eye-wear recommender system. Our approach is based on extracting relevant facial landmark measurements and passing them through a naive Bayes classifier unit in order to yield the final decision. The literature on this subject is particularly scarce owing to the very subjective nature of human face shape classification. We wish to contribute a robust and automatic system that performs this task and highlight future development directions on this matter.

Central Object Segmentation by Deep Learning to Continuously Monitor Fruit Growth through RGB Images

Monitoring fruit growth is useful when estimating final yields in advance and predicting optimum harvest times. However, observing fruit all day at the farm via RGB images is not an easy task because the light conditions are constantly changing. In this paper, we present CROP (Central Roundish Object Painter). The method involves image segmentation by deep learning, and the architecture of the neural network is a deeper version of U-Net. CROP identifies different types of central roundish fruit in an RGB image in varied light conditions, and creates a corresponding mask. Counting the mask pixels gives the relative two-dimensional size of the fruit, and in this way, time-series images may provide a non-contact means of automatically monitoring fruit growth. Although our measurement unit is different from the traditional one (length), we believe that shape identification potentially provides more information. Interestingly, CROP can have a more general use, working even for some other roundish objects. For this reason, we hope that CROP and our methodology yield big data to promote scientific advancements in horticultural science and other fields.

Inverse Problem Related to Boundary Shape Identification for a Hyperbolic Differential Equation

In this paper, we are interested in the inverse problem of the determination of the unknown part ∂ Ω , Γ 0 of the boundary of a uniformly Lipschitzian domain Ω included in ℝ N from the measurement of the normal derivative ∂ n v on suitable part Γ 0 of its boundary, where v is the solution of the wave equation ∂ t t v x , t − Δ v x , t + p x v x = 0 in Ω × 0 , T and given Dirichlet boundary data. We use shape optimization tools to retrieve the boundary part Γ of ∂ Ω . From necessary conditions, we estimate a Lagrange multiplier k Ω which appears by derivation with respect to the domain. By maximum principle theory for hyperbolic equations and under geometrical assumptions, we prove a uniqueness result of our inverse problem. The Lipschitz stability is established by increasing of the energy of the system. Some numerical simulations are made to illustrate the optimal shape.

Development and testing of a novel image analysis algorithm for descriptive evaluation of shape change of a shrinkable soft material

Soft material can undergo non-uniform deformation or change of shape upon processing. Identifying shape and its change is nevertheless not straightforward. In this study, novel image-based algorithm that can be used to identify shapes of input images and at the same time classify non-uniform deformation into various patterns, i.e., swelling/shrinkage, horizontal and vertical elongations/contractions as well as convexity and concavity, is proposed. The algorithm was first tested with computer-generated images and later applied to agar cubes, which were used as model shrinkable soft material, undergoing drying at different temperatures. Shape parameters and shape-parameter based algorithm as well as convolutional neural networks (CNNs) either incorrectly identified some complicated shapes or could only identify the point where non-uniform deformation started to take place; CNNs lacked ability to describe non-uniform deformation evolution. Shape identification accuracy of the newly developed algorithm against computer-generated images was 65.88%, while those of the other tested algorithms ranged from 34.76 to 97.88%. However, when being applied to the deformation of agar cubes, the developed algorithm performed superiorly to the others. The proposed algorithm could both identify the shapes and describe their changes. The interpretation agreed well with that via visual observation.

Extraction and Classification of Mouth Shape Features in Oral English Teaching Based on Image Processing

Mouth shape identification helps oral English learners discover the features of their lip movements in English speaking, and correct their pronunciation more smoothly. So far, few scholars have applied image processing to identify mouth shape features of oral English learners. Most studies consider little about environmental factors, and ignore the changing mouth shape in pronunciation. Therefore, this paper explores the extraction and classification of mouth shape features in oral English teaching based on image processing. Firstly, an extraction and classification model were established for mouth shape features in oral English teaching. Then, the mouth shape images of oral English teaching were preprocessed. After that, the authors segmented the lips in oral English video frames based on neural network, extracted the lip boundaries from the said frames, and fitted them into curves. The proposed model was proved effective through experiments.

AN EFFICIENT IDENTIFICATION OF RED BLOOD CELL EQUILIBRIUM SHAPE USING NEURAL NETWORKS

This work of applied mathematics with interfaces in bio-physics focuses on the shape identification and numerical modelisation of a single red blood cell shape. The purpose of this work is to provide a quantitative method for interpreting experimental observations of the red blood cell shape under microscopy. In this paper we give a new formulation based on classical theory of geometric shape minimization which assumes that the curvature energy with additional constraints controls the shape of the red blood cell. To minimize this energy under volume and area constraints, we propose a new hybrid algorithm which combines Particle Swarm Optimization (PSO), Gravitational Search (GSA) and Neural Network Algorithm (NNA). The results obtained using this new algorithm agree well with the experimental results given by Evans et al. (8) especially for sphered and biconcave shapes.

Real time Adaptive Approach for Hidden Targets Shape Identification using through Wall Imaging System

In Through-wall Imaging (TWI) system, shape-based identification of the hidden target behind the wall made of any dielectric material like brick, cement, concrete, dry plywood, plastic and Teflon, etc. is one of the most challenging tasks. However, it is very important to understand that the performance of TWI systems is limited by the presence of clutter due to the wall and also transmitted frequency range. Therefore, the quality of obtained image is blurred and very difficult to identify the shape of targets. In the present paper, a shape-based image identification technique with the help of a neural network and curve-fitting approach is proposed to overcome the limitation of existing techniques. A real time experimental analysis of TWI has been carried out using the TWI radar system to collect and process the data, with and without targets. The collected data is trained by a neural network for shape identification of targets behind the wall in any orientation and then threshold by a curve-fitting method for smoothing the background. The neural network has been used to train the noisy data i.e. raw data and noise free data i.e. pre-processed data. The shape of hidden targets is identified by using the curve fitting method with the help of trained neural network data and real time data. The results obtained by the developed technique are promising for target identification at any orientation.