Video Sequence Analysis for On-Table Tennis Player Ranking and Analysis

With the rapidly developing of the scientific research in the field of sports, big data analytics and information science are used to carry out technical and tactical statistical analysis of competition or training videos. The table tennis is a skill oriented sport. The technique and tactics in table tennis are the core factors to win the game. With the endlessly emerging innovative playing techniques and tactics, the players have their own competition styles. According to the competition events among athletes, the athletes’ competition relationship network is constructed and the players’ ranking is established. The ranking can be used to help table tennis players improve daily training and understand their ability. In this paper, the table tennis players’ ranking is established their competition videos and their prestige scores in the table tennis players’ competition relationship network.

Using Deep Learning for Collecting Data about Museum Visitor Behavior

Nowadays, technology makes it possible to admire objects and artworks exhibited all over the world remotely. We have been able to appreciate this convenience even more in the last period, in which the pandemic has forced us into our homes for a long time. However, visiting art sites in person remains a truly unique experience. Even during on-site visits, technology can help make them much more satisfactory, by assisting visitors during the fruition of cultural and artistic resources. To this aim, it is necessary to monitor the active user for acquiring information about their behavior. We, therefore, need systems able to monitor and analyze visitor behavior. The literature proposes several techniques for the timing and tracking of museum visitors. In this article, we propose a novel approach to indoor tracking that can represent a promising and non-expensive solution for some of the critical issues that remain. In particular, the system we propose relies on low-cost equipment (i.e., simple badges and off-the-shelf RGB cameras) and harnesses one of the most recent deep neural networks (i.e., Faster R-CNN) for detecting specific objects in an image or a video sequence with high accuracy. An experimental evaluation performed in a real scenario, namely, the “Exhibition of Fake Art” at Roma Tre University, allowed us to test our system on site. The collected data has proven to be accurate and helpful for gathering insightful information on visitor behavior.

Assistive Technology based Solution for Hearing Impairment using Smartphone

Developing a system for sign language recognition becomes essential for the deaf as well as a mute person. The recognition system acts as a translator between a disabled and an able person. This eliminates the hindrances in the exchange of ideas. Most of the existing systems are very poorly designed with limited support for the needs of their day to day facilities. The proposed system embedded with gesture recognition capability has been introduced here which extracts signs from a video sequence and displays them on screen. On the other hand, a speech to text as well as text to speech system is also introduced to further facilitate the grieved people. To get the best out of a human-computer relationship, the proposed solution consists of various cutting-edge technologies and Machine Learning based sign recognition models that have been trained by using TensorFlow and Keras library. The proposed architecture works better than several gesture recognition techniques like background elimination and conversion to HSV

Devising a method for measuring the motion parameters of industrial equipment in the quarry using adaptive parameters of a video sequence

The method and structural scheme of an information-measuring system for determining the parameters of objects' movements (technological equipment in the quarry for extracting block natural stone) have been proposed. A distinctive feature of time video sequences containing images of measured objects is their adaptation and adjustment in accordance with the intensity of movement and accuracy requirements for measurement results. Structural and software-algorithmic methods were also applied for improving the accuracy of measurements of motion parameters, namely: complexation of two measuring channels and exponential smoothing of digital references. One of the measuring channels is based on a digital video camera, the second is based on an accelerometer mounted on an object and two integrators. Exponential smoothing makes it possible to take into consideration the previous countdowns of movement parameters with weight coefficients. That ensures accounting for the existing patterns of movement of the object and reducing the errors when measuring the parameters of movement by (1.4...1.6) times. The resulting solutions have been implemented in the form of an information and measurement system. The technological process of extracting blocks of natural stone in the quarry was experimentally investigated using a diamond-rope installation. Based on the contactless measurement of motion parameters, it is possible to ensure control over this process and improve the quality of blocks made of natural stone. Based on the experimental study of measurement errors, recommendations were given for the selection of adaptive parameters of a video sequence, namely the size of images and the value of the inter-frame interval. In addition, methods for the software-algorithmic processing of measuring information were selected, specifically exponential smoothing and averaging the coordinates of the contour of an object, measured in 30 adjacent lines of the image

Researching a machine learning algorithm for a face recognition system

This article investigated the problem of using machine learning algorithms to recognize and identify a user in a video sequence. The scientific novelty lies in the proposed improved Viola-Jones method, which will allow more efficient and faster recognition of a person's face. The practical value of the results obtained in the work is determined by the possibility of using the proposed method to create systems for human face recognition. A review of existing methods of face recognition, their main characteristics, architecture and features was carried out. Based on the study of methods and algorithms for finding faces in images, the Viola-Jones method, wavelet transform and the method of principal components were chosen. These methods are among the best in terms of the ratio of recognition efficiency and work speed. Possible modifications of the Viola-Jones method are presented. The main contribution presented in this article is an experimental study of the impact of various types of noise and the improvement of company security through the development of a computer system for recognizing and identifying users in a video sequence. During the study, the following tasks were solved: – a model of face recognition is proposed, that is, the system automatically detects a person's face in the image (scanned photos or video materials); – an algorithm for analyzing a face is proposed, that is, a representation of a person's face in the form of 68 modal points; – an algorithm for creating a digital fingerprint of a face, which converts the results of facial analysis into a digital code; – development of a match search module, that is, the module compares the faceprint with the database until a match is found

Standardized Judgment Method of Shooting Training Action Based on Digital Video Technology

Aiming at the difficulty of standardizing the action of basketball shooting training, a new method of standardizing the action of basketball shooting training is proposed based on digital video technology. The digital video signal representation, video sequence coding data structure, and video sequence compression coding method are analyzed, and the pixels of basketball shooting training action position space are sampled to collect basketball shooting training images. The time difference method is used to extract the movement target of basketball shooting training from a digital video sequence. Based on digital video technology, the initial background image is estimated, and the update rate is introduced to update the background estimation image. According to the pixel value sequence of the basketball shooting training image, the pixel model of the basketball shooting training image is defined and modified. By judging whether the defined pixel value matches the background parameter model, the standardization of shooting training can be realized. The experimental results show that the proposed method has good stability, high precision, and short time in determining the standardization of shooting movement, can correct the wrong shooting movement in real time, and can effectively guide basketball shooting training.

RESEARCH INFORMATION SYSTEM OF IMAGE RECOGNITION

Data classification in presence of noise can lead to much worse results than expected for pure patterns. In paper was investigated problem of the research is the process of user recognition and identification in the video sequence. The main contributions presented in this paper are experimental examination of influence of different types of noise and to the increase the security of an IT company by developing a computer system for recognizing and identifying users in the video sequence. Based on the study of methods and algorithms for finding faces in images, the Viola-Jones method, wavelet transform and the method of principal components were chosen. These methods are among the best in terms of the ratio of recognition efficiency and work speed. However, the training of classifiers is very slow, but the face search results are very fast.

Monocular weakly supervised depth and pose estimation method based on multi-information fusion

The depth and pose information are the basic issues in the field of robotics, autonomous driving, and virtual reality, and are also the focus and difficult issues of computer vision research. The supervised monocular depth and pose estimation learning are not feasible in environments where labeled data is not abundant. Self-supervised monocular video methods can learn effectively only by applying photometric constraints without expensive ground true depth label constraints, which results in an inefficient training process and suboptimal estimation accuracy. To solve these problems, a monocular weakly supervised depth and pose estimation method based on multi-information fusion is proposed in this paper. First, we design a high-precision stereo matching method to generate a depth and pose data as the "Ground Truth" labels to solve the problem that the ground truth labels are difficult to obtain. Then, we construct a multi-information fusion network model based on the "Ground truth" labels, video sequence, and IMU information to improve the estimation accuracy. Finally, we design the loss function of supervised cues based on "Ground Truth" labels cues and self-supervised cues to optimize our model. In the testing phase, the network model can separately output high-precision depth and pose data from a monocular video sequence. The resulting model outperforms mainstream monocular depth and poses estimation methods as well as the partial stereo matching method in the challenging KITTI dataset by only using a small number of real training data(200 pairs).

Facial Expression Recognition from Video Sequence Using Self Organizing Feature Map

Automatic analysis of facial expressions is rapidly becoming an area of intense interest in computer vision and artificial intelligence research communities. In this paper an approach is presented for facial expression recognition of the six basic prototype expressions (i.e., joy, surprise, anger, sadness, fear, and disgust) based on Facial Action Coding System (FACS). The approach utilizes the topological ordering patterns produced by Kohonen Self Organizing Map, in which implemented on expression image sequence for each prototype facial expression. The map will compute the topological relationship between the particular expression sequences, starting from the neutral expression to peak. This method tried to find a topological ordering pattern (shape) for each expression; it will not require any pre-processing tedious work such as normalization. The only requirement is that, image background must be kept constant, also with non-rigid head motion. The feature extraction phase had been performed by this method, while the classification phase done by especially designed procedures for shape and direction finding to recognize the pattern of the shape, thereafter the type of the expression also backpropagation neural network is implemented for the classification task. An average recognition rate of 88.7% was achieved for six basic expressions, where different databases had been used for the test of the method.

Computer simulation of adapting the object image selection method to video surveillance conditions

The article deals with the issues of computer modeling of methods for selecting images of objects against a non-uniform background. A test video sequence with given background and object parameters is considered, which provides imitation of one of the special cases of video surveillance conditions, namely, the convergence of the video surveillance point and the object. The issues of adaptation of the compensation selection method to the specified conditions of video surveillance are discussed. Examples of test images and graphs of dependences of the probability of correct determination of coordinates depending on the value of the local contrast of the object in relation to the background, obtained by computer simulation are given.