Chessboard and Chess Piece Recognition With the Support of Neural Networks

AbstractChessboard and chess piece recognition is a computer vision problem that has not yet been efficiently solved. Digitization of a chess game state from a picture of a chessboard is a task typically performed by humans or with the aid of specialized chessboards and pieces. However, those solutions are neither easy nor convenient. To solve this problem, we propose a novel algorithm for digitizing chessboard configurations.We designed a method of chessboard recognition and pieces detection that is resistant to lighting conditions and the angle at which images are captured, and works correctly with numerous chessboard styles. Detecting the board and recognizing chess pieces are crucial steps of board state digitization.The algorithm achieves 95% accuracy (compared to 60% for the best alternative) for positioning the chessboard in an image, and almost 95% for chess pieces recognition. Furthermore, the sub-process of detecting straight lines and finding lattice points performs extraordinarily well, achieving over 99.5% accuracy (compared to the 74% for the best alternative).

Download Full-text

Reconstruction Lighting Positions of a Scene for Mixed Reality Systems Using Convolutional Neural Network and Shadow Ray Tracing

Keldysh Institute Preprints ◽

10.20948/prepr-2021-67 ◽

2021 ◽

pp. 1-18

Author(s):

Maxim Igorevich Sorokin ◽

Dmitry Dmitrievich Zhdanov ◽

Ildar Vagizovich Valiev

Keyword(s):

Neural Network ◽

Neural Networks ◽

Computer Vision ◽

Ray Tracing ◽

Light Source ◽

Virtual Worlds ◽

Mixed Reality ◽

The Real ◽

Virtual Objects ◽

Lighting Conditions

The paper examines the causes of visual discomfort in mixed reality systems and algorithmic solutions that eliminate one of the main causes of discomfort, namely, the mismatch between the lighting conditions of objects in the real and virtual worlds. To eliminate this cause of discomfort, the algorithm is proposed, which consists in constructing groups of shadow rays from points on the boundaries of shadows to points on the boundaries of objects. Part of the rays corresponding to the real lighting conditions form caustics in area of the real light source, which makes it possible to determine the source of illumination of virtual objects for their correct embedding into the mixed reality system. Convolutional neural networks and computer vision algorithms were used to classify shadows in the image. Examples of reconstructing the coordinates of a light source from RGBD data are presented.

Download Full-text

Determining Chess Game State from an Image

Journal of Imaging ◽

10.3390/jimaging7060094 ◽

2021 ◽

Vol 7 (6) ◽

pp. 94

Author(s):

Georg Wölflein ◽

Ognjen Arandjelović

Keyword(s):

Computer Vision ◽

Projective Transformation ◽

Recognition System ◽

Inference System ◽

Starting Position ◽

Current State ◽

Chess Game ◽

Order Of Magnitude ◽

Game State ◽

Chess Players

Identifying the configuration of chess pieces from an image of a chessboard is a problem in computer vision that has not yet been solved accurately. However, it is important for helping amateur chess players improve their games by facilitating automatic computer analysis without the overhead of manually entering the pieces. Current approaches are limited by the lack of large datasets and are not designed to adapt to unseen chess sets. This paper puts forth a new dataset synthesised from a 3D model that is an order of magnitude larger than existing ones. Trained on this dataset, a novel end-to-end chess recognition system is presented that combines traditional computer vision techniques with deep learning. It localises the chessboard using a RANSAC-based algorithm that computes a projective transformation of the board onto a regular grid. Using two convolutional neural networks, it then predicts an occupancy mask for the squares in the warped image and finally classifies the pieces. The described system achieves an error rate of 0.23% per square on the test set, 28 times better than the current state of the art. Further, a few-shot transfer learning approach is developed that is able to adapt the inference system to a previously unseen chess set using just two photos of the starting position, obtaining a per-square accuracy of 99.83% on images of that new chess set. The code, dataset, and trained models are made available online.

Download Full-text

The machine training in problems of satellite images’s processing

Metrologiya ◽

10.32446/0132-4713.2020-4-15-37 ◽

2020 ◽

pp. 15-37

Author(s):

L. P. Bass ◽

Yu. A. Plastinin ◽

I. Yu. Skryabysheva

Keyword(s):

Remote Sensing ◽

Neural Networks ◽

Computer Vision ◽

Satellite Images ◽

Vision Systems ◽

Earth Remote Sensing ◽

Practical Applications ◽

Convolution Neural Networks ◽

Computer Vision Systems ◽

Trained Neural Network

Use of the technical (computer) vision systems for Earth remote sensing is considered. An overview of software and hardware used in computer vision systems for processing satellite images is submitted. Algorithmic methods of the data processing with use of the trained neural network are described. Examples of the algorithmic processing of satellite images by means of artificial convolution neural networks are given. Ways of accuracy increase of satellite images recognition are defined. Practical applications of convolution neural networks onboard microsatellites for Earth remote sensing are presented.

Download Full-text

Algorithms for segmentation and recognition of objects on medical images based on chiarlet transformation and neural networks

Informatization and communication ◽

10.34219/2078-8320-2020-11-2-35-45 ◽

2020 ◽

pp. 35-45

Author(s):

Y.A. Hamad ◽

K.V. Simonov ◽

A.S. Kents

Keyword(s):

Neural Networks ◽

Image Processing ◽

Computer Vision ◽

Edge Detection ◽

Medical Images ◽

Classification Algorithms ◽

Visual Data ◽

Recognition Of Objects

The paper considers general approaches to image processing, analysis of visual data and computer vision. The main methods for detecting features and edges associated with these approaches are presented. A brief description of modern edge detection and classification algorithms suitable for isolating and characterizing the type of pathology in the lungs in medical images is also given.

Download Full-text

Developing Algorithms for a Berry Recognition System Used in Robotized Harvesting of Garden Strawberry

Elektrotekhnologii i elektrooborudovanie v APK ◽

10.22314/2658-4859-2020-67-1-133-141 ◽

2020 ◽

Vol 67 (1) ◽

pp. 133-141

Author(s):

Dmitriy O. Khort ◽

Aleksei I. Kutyrev ◽

Igor G. Smirnov ◽

Rostislav A. Filippov ◽

Roman V. Vershinin

Keyword(s):

Computer Vision ◽

Technological Process ◽

Degrees Of Freedom ◽

Computer Control ◽

Recognition System ◽

Raspberry Pi ◽

Laser Sensors ◽

Lighting Conditions ◽

Systems Research ◽

Degree Of Maturity

Technological capabilities of agricultural units cannot be optimally used without extensive automation of production processes and the use of advanced computer control systems. (Research purpose) To develop an algorithm for recognizing the coordinates of the location and ripeness of garden strawberries in different lighting conditions and describe the technological process of its harvesting in field conditions using a robotic actuator mounted on a self-propelled platform. (Materials and methods) The authors have developed a self-propelled platform with an automatic actuator for harvesting garden strawberry, which includes an actuator with six degrees of freedom, a co-axial gripper, mg966r servos, a PCA9685 controller, a Logitech HD C270 computer vision camera, a single-board Raspberry Pi 3 Model B+ computer, VL53L0X laser sensors, a SZBK07 300W voltage regulator, a Hubsan X4 Pro H109S Li-polymer battery. (Results and discussion) Using the Python programming language 3.7.2, the authors have developed a control algorithm for the automatic actuator, including operations to determine the X and Y coordinates of berries, their degree of maturity, as well as to calculate the distance to berries. It has been found that the effectiveness of detecting berries, their area and boundaries with a camera and the OpenCV library at the illumination of 300 Lux reaches 94.6 percent’s. With an increase in the robotic platform speed to 1.5 kilometre per hour and at the illumination of 300 Lux, the average area of the recognized berries decreased by 9 percent’s to 95.1 square centimeter, at the illumination of 200 Lux, the area of recognized berries decreased by 17.8 percent’s to 88 square centimeter, and at the illumination of 100 Lux, the area of recognized berries decreased by 36.4 percent’s to 76 square centimeter as compared to the real area of berries. (Conclusions) The authors have provided rationale for the technological process and developed an algorithm for harvesting garden strawberry using a robotic actuator mounted on a self-propelled platform. It has been proved that lighting conditions have a significant impact on the determination of the area, boundaries and ripeness of berries using a computer vision camera.

Download Full-text

Bipolar Morphological Neural Networks: Gate-Efficient Architecture for Computer Vision

IEEE Access ◽

10.1109/access.2021.3094484 ◽

2021 ◽

pp. 1-1

Author(s):

Elena E. Limonova ◽

Daniil M. Alfonso ◽

Dmitry P. Nikolaev ◽

Vladimir V. Arlazarov

Keyword(s):

Neural Networks ◽

Computer Vision ◽

Morphological Neural Networks

Download Full-text

Off-the-shelf deep learning is not enough, and requires parsimony, Bayesianity, and causality

npj Computational Materials ◽

10.1038/s41524-020-00487-0 ◽

2021 ◽

Vol 7 (1) ◽

Author(s):

Rama K. Vasudevan ◽

Maxim Ziatdinov ◽

Lukas Vlcek ◽

Sergei V. Kalinin

Keyword(s):

Neural Networks ◽

Computer Vision ◽

Deep Learning ◽

Bayesian Methods ◽

Deep Neural Networks ◽

Applied Research ◽

Modern Science ◽

Generative Models ◽

Knowledge Development ◽

Physical Constraints

AbstractDeep neural networks (‘deep learning’) have emerged as a technology of choice to tackle problems in speech recognition, computer vision, finance, etc. However, adoption of deep learning in physical domains brings substantial challenges stemming from the correlative nature of deep learning methods compared to the causal, hypothesis driven nature of modern science. We argue that the broad adoption of Bayesian methods incorporating prior knowledge, development of solutions with incorporated physical constraints and parsimonious structural descriptors and generative models, and ultimately adoption of causal models, offers a path forward for fundamental and applied research.

Download Full-text

Deep learning-enabled medical computer vision

npj Digital Medicine ◽

10.1038/s41746-020-00376-2 ◽

2021 ◽

Vol 4 (1) ◽

Author(s):

Andre Esteva ◽

Katherine Chou ◽

Serena Yeung ◽

Nikhil Naik ◽

Ali Madani ◽

...

Keyword(s):

Artificial Intelligence ◽

Neural Networks ◽

Computer Vision ◽

Deep Learning ◽

Medical Imaging ◽

Real World ◽

Recent Progress ◽

Medical Applications ◽

Modern Computer ◽

Medical Computer

AbstractA decade of unprecedented progress in artificial intelligence (AI) has demonstrated the potential for many fields—including medicine—to benefit from the insights that AI techniques can extract from data. Here we survey recent progress in the development of modern computer vision techniques—powered by deep learning—for medical applications, focusing on medical imaging, medical video, and clinical deployment. We start by briefly summarizing a decade of progress in convolutional neural networks, including the vision tasks they enable, in the context of healthcare. Next, we discuss several example medical imaging applications that stand to benefit—including cardiology, pathology, dermatology, ophthalmology–and propose new avenues for continued work. We then expand into general medical video, highlighting ways in which clinical workflows can integrate computer vision to enhance care. Finally, we discuss the challenges and hurdles required for real-world clinical deployment of these technologies.

Download Full-text

Image classification using Deep learning

International Journal of Engineering & Technology ◽

10.14419/ijet.v7i2.7.10892 ◽

2018 ◽

Vol 7 (2.7) ◽

pp. 614 ◽

Cited By ~ 5

Author(s):

M Manoj krishna ◽

M Neelima ◽

M Harshali ◽

M Venu Gopala Rao

Keyword(s):

Machine Learning ◽

Neural Networks ◽

Image Processing ◽

Computer Vision ◽

Deep Learning ◽

Image Classification ◽

Convolutional Neural Networks ◽

Classical Problem

The image classification is a classical problem of image processing, computer vision and machine learning fields. In this paper we study the image classification using deep learning. We use AlexNet architecture with convolutional neural networks for this purpose. Four test images are selected from the ImageNet database for the classification purpose. We cropped the images for various portion areas and conducted experiments. The results show the effectiveness of deep learning based image classification using AlexNet.

Download Full-text