Computer Vision

Smart coaching and e-sport platforms have shown a great interest in the recent research studies. Through this study, we aim to globalize the practice of sport, especially Shotokan Karate, by connecting participants and coaches on an international scale through the integration of Artificial Intelligence techniques such as computer vision and deep learning, to give the possibility of carrying out national and international virtual training courses without logistical constraints. The proposed work aims to apply the latest action detection, action recognition, and action classification methods for different Karate movements using the LSTM and the ST-GCN algorithms and proposes these movements in 3D using Video Inference for Human Body Pose and Shape Estimation (VIBE). Our proper Multiview Dataset contains pose estimations of a set of basic movements captured by a karate Shotokan expert (6th DAN Black Belt) from three views (Front view, Left view, and Right view) using OpenPose and FastP...

Surveying Madrid Air Pollution is a challenging problem in deception detection, and it has tremendous real-world political and social impacts. Ozone and PM10 constitute the major concern for air quality of Madrid. This paper addresses the problem of the prediction of such two pollutants, using to this end several statistical approaches. In particular, feed-forward neural networks (FFNNs), currently recognized as state-of-the-art approach for statistical prediction of air quality, and are compared with two alternative approaches derived from machine learning: pruned neural networks (PNNs) and lazy learning (LL). PNNs constitute a parameter-parsimonious approach, based on the removal of redundant parameters from fully connected neural networks; LL, on the other hand, is a local linear prediction algorithm, which performs a local learning procedure each time a prediction is required. The global environment is currently facing a major issue of air pollution. It is one of the foremost ca...

In content-based 3D object retrieval, searching for a query object in an extensive database is essential. The existing retrieval algorithms adopt the naïve search algorithm in searching for query objects. This approach leads to a high cost of search and retrieval that needs to be addressed. In this research, we introduced an algorithm that calculates each cluster’s representative, that is, a 3D object that has the least dissimilarity on average to each object in the cluster. This is to improve the overall retrieval performance of [1]. We first compute the optimal hierarchical level of the database using a dendrogram, and then calculate the total number of clusters in the database. Afterwards, we calculate the feature descriptor of each cluster. When a user chooses a query object, our system then compares the feature descriptor of the query object with each of the cluster’s representation and search the cluster with the smallest distance to the query, thereby improving the query sear...

The intelligent environments, built upon many distributed sensors, are promising technology for ubiquitous interaction between robots and human beings. Especially, it is important to track target objects and get the positional information of them in such environments. This paper focuses on adapting camera selection for target tracking in multi-camera system. In this paper, a fuzzy automaton based camera selection method is introduced. In the proposed method, the camera selection decision is driven by fuzzy automaton based on the previously selected camera (previous camera state) and the tracking level of the object in each available camera. Simulations for evaluation of the proposed method and comparison with the previous method are presented. The results show that the proposed method is efficient for adaptive camera selection in multi-camera environment and helps easy construction of multi-camera placement. An actual multi-camera system with the proposed camera selection method was...

Smart city infrastructure and Intelligent Transportation Systems (ITS) need modern traffic monitoring and driver assistance systems such as autonomous traffic signal detection. ITS is a dominant research area among several fields in the domain of artificial intelligence. Traffic signal detection is a key module of autonomous vehicles where accuracy and inference time are amongst the most significant parameters. In this regard, the aim of this study is to detect traffic signals focusing to enhance accuracy and real-time performance. The results and discussion enclose a comparative performance of a CNN-based algorithm YOLO V3 and a handcrafted technique that gives insight for enhanced detection and inference in day and night light. It is important to consider that real-world objects are associated with complex backgrounds, occlusion, climate conditions, and light exposure that deteriorate the performance of sensitive intelligent applications. This study provides a direction to propose...

In this paper, we present the first realisation and experimentation of a new eye tracking system using an infrared (iR) laser pointer embedded into a wireless smart contact lens. We denote this contact lens prototype as the cyclops lens, in reference to the famous hero of the X-Men comics. The full eye tracker device combines the smart contact lens and its eyewear, which provides a primary source of energy and the beam detection system. We detail the assembling and encapsulation process of the main functionalities into the contact lens and present how a gaze tracking system is achieved, compared to existing conventional eye-tracking ones. Finally, we discuss future technical improvements.

This study aims to apply and analyze Indonesian Part of Speech Tagging (POS Tagging) using the Decision Tree algorithm. The POS Tagging process can use language dictionaries as a reference or assist in the provision of word classes (tags), but there are some constraints such as the determination of word classes for ambiguous words and words that are not in the dictionary, so a further POS Tagging approach is needed to overcome this problem. This study uses Indonesian Language Dictionary (KBBI) data and Indonesian CNN online news data. The news data is then carried out the process of preprocessing, feature extraction, and vectorization so as to produce a dataset of 40,071 which are then used as training data to build the Decision Tree model. POS Tagging testing in this study obtained an accuracy of 95.3%, and a percentage of success of 90.0% in tagging words that were not on KBBI and success of 91.9% for tagging ambiguous words. The results of the implementation of the POS Tagging ap...

Measure the risk of developing pressure injuries in the elderly with immobility and investigate possible preventive measures. Cross-sectional, descriptive study with a quantitative approach. Sample of 50 elderly people evaluated using the Braden scale and, subsequently, the occurrence of preventive measures was investigated. The data were processed using the “R” interface software, using the Mann-Whitney statistical tests and using the hierarchical mean link method. A strong influence of the variable age and education related to prevention measures was identified. Regarding the risk of developing pressure injuries, the elderly fall into the “High Risk” category. It was observed the use of clean sheets and free of residues 49 (98%) and the use of pillows, cushions and similar 45 (90%). The elderly are at high risk for the development of pressure injuries and preventive measures are proven to be effective in preventing such comorbidity.

Automatic image annotation at region-level consists of the task of assigning labels to regions within segmented images. This is a very important task for the development and improvement of image retrieval methods. Many image annotation methods have been proposed so far, reporting relatively good results on this task. However the lack of benchmark collections have caused that region-level methods are often evaluated using image-level collections. On the other hand, a few region-level collections are available, ...

This coronary-artery extraction method uses one initialization point per vessel. First, a mask is computed by use of a region-growing algorithm, which starts from the initial point and stops when no more connected voxels fall within an interactively defined intensity range. The centerline tracking is then performed within the mask, starting from the same initial point. This algorithm is based on a prediction/estimation scheme. It uses the first- and second-order image moments calculated within a spherical volume that slides along the vessel, and the radius of which is automatically adjusted to the local radius of the vessel. The evolution of the radius of the sphere is based on the analysis of the eigenvalues of the inertia matrix in a multi-scale framework. The estimation of the current point location makes use of an elastic model similar to ”snakes”. The point iteratively moves under the action of an image-force attracting it to the local gravity center, and under the reaction of ...