An Efficient 3D Human Pose Retrieval and Reconstruction from 2D Image-Based Landmarks

We propose an efficient and novel architecture for 3D articulated human pose retrieval and reconstruction from 2D landmarks extracted from a 2D synthetic image, an annotated 2D image, an in-the-wild real RGB image or even a hand-drawn sketch. Given 2D joint positions in a single image, we devise a data-driven framework to infer the corresponding 3D human pose. To this end, we first normalize 3D human poses from Motion Capture (MoCap) dataset by eliminating translation, orientation, and the skeleton size discrepancies from the poses and then build a knowledge-base by projecting a subset of joints of the normalized 3D poses onto 2D image-planes by fully exploiting a variety of virtual cameras. With this approach, we not only transform 3D pose space to the normalized 2D pose space but also resolve the 2D-3D cross-domain retrieval task efficiently. The proposed architecture searches for poses from a MoCap dataset that are near to a given 2D query pose in a definite feature space made up of specific joint sets. These retrieved poses are then used to construct a weak perspective camera and a final 3D posture under the camera model that minimizes the reconstruction error. To estimate unknown camera parameters, we introduce a nonlinear, two-fold method. We exploit the retrieved similar poses and the viewing directions at which the MoCap dataset was sampled to minimize the projection error. Finally, we evaluate our approach thoroughly on a large number of heterogeneous 2D examples generated synthetically, 2D images with ground-truth, a variety of real in-the-wild internet images, and a proof of concept using 2D hand-drawn sketches of human poses. We conduct a pool of experiments to perform a quantitative study on PARSE dataset. We also show that the proposed system yields competitive, convincing results in comparison to other state-of-the-art methods.

Hyperbolic Medicine: General and Local Hyperbolic Curves on Earth, Influencing Human Physiology

The objective of this work is to study the hyperbolic space-time curves that we perceive in our daily lives through images obtained in different geographical locations on Earth. A bibliographic review of scientific works has been made on hyperbolic curves in medicine, electromagnetic fields, circadian rhythms, and space-time perpendicular to the movement of an organ. On the internet, images in conical perspective have been selected, of cities located at different longitudes and latitudes within the geography of the Earth. They have tried to identify hyperbolic space-time curves. They have then been compared with the hyperbolic curves of the Earth’s magnetic field. Finally, we determine its relationship with the hyperbolic curves of human physiology. We have as conclusions: 1. Human vision is hyperbolic because the space in which we live is deformed by “general hyperbolic curves” that exist at any longitude and latitude of the earth’s geography. 2. The lines of force of the Earth’s magnetic field are hyperbolas that follow “local hyperbolic curves” since they can vary in intensity and even reverse their polarity over time. 3. In the physiology of the human body there are hyperbolic curves that are similar to the lines of force of a magnet and the magnetic field of the Earth. Human physiology can be conditioned by general hyperbolic curves and by local hyperbolic curves. There is an adaptation to that hyperbolic deformation of the space in which we live.

Narrative as a Radial Category

Narrative is a complex and elusive category of cognition, culture, communication and language. An attempt has been made in this article with a large enough theoretical scope to consider the possibility of treating narrative as a radial category. To this end, the definition and characterisation of radiality is provided together with explanation of what it might mean to apply this term to the complex language-discourse unit of narrative. The prototype of this category involves features, functions, and ICMs. It has multiple representations with only family resemblance, involves more obvious exemplars and variable abstract knowledge structures. In particular, section one looks at the radiality question and what it might mean to think of the meaning of narrative in general. Section two focuses on centrality. Sections three to five deal with schematic representations of narrative and provide examples of extending the most subsuming schema of the Action Chain Model from cognitive linguistics and Labov’s Narrative Schema to various other types of conversational narrative, children’s dramatic plays, tactical narratives, story rounds, jokes, poems, current news articles on the Internet, images, and advertisements.

Human Gender Classification Using Transfer Learning via Pareto Frontier CNN Networks

Human gender is deemed as a prime demographic trait due to its various usage in the practical domain. Human gender classification in an unconstrained environment is a sophisticated task due to large variations in the image scenarios. Due to the multifariousness of internet images, the classification accuracy suffers from traditional machine learning methods. The aim of this research is to streamline the gender classification process using the transfer learning concept. This research proposes a framework that performs automatic gender classification in unconstrained internet images deploying Pareto frontier deep learning networks; GoogleNet, SqueezeNet, and ResNet50. We analyze the experiment with three different Pareto frontier Convolutional Neural Network (CNN) models pre-trained on ImageNet. The massive experiments demonstrate that the performance of the Pareto frontier CNN networks is remarkable in the unconstrained internet image dataset as well as in the frontal images that pave the way to developing an automatic gender classification system.

A Dwt and Scrambling Based Blind Watermarking Algorithm for Digital Images Authentication

Like the other multimedia that is spread on the Internet, images are also vulnerable to theft and attacks. Protecting the image is therefore an urgent necessity because it represents a large proportion of the digital content. Authentication and ownership protection are the basic demands of image security and these are achieved by applying watermarking techniques. For the Muslim world, the Holy Quran has its sanctity, which does not accept any controversy or doubt. As part of keeping pace with modern technology, digital copies of the Holy Qur’an are available, which are widely distributed all over the world. Therefore, it is necessary to ensure that these copies maintain their integrity and ensure that there are no malicious manipulations. In this paper, we propose an image watermarking scheme to authenticate the images of digital version of Holy Quran using discrete wavelet transform DWT. Here a fragile watermark is used to clarify whether there is any modification occurred to the intended images. Initially the cover image is decomposed by DWT where 2nd and 4th level coefficients are exploited for watermark embedding. The intended watermark is obtained by scrambling the original cover image. Then the scrambled image is inserted into the DWT coefficients by several trials using different embedding gains. To evaluate our system and see how effective it is to detect any error or manipulation, PSNR, SSIM and MSE are employed beside that they are acting as an imperceptibility measure. Results proved that our method has achieved a good level of imperceptibility and can detect any slight tamper. It is necessary to bear in mind that this method is valid for application to normal color images as well and gives an excellent level of efficiency