Reliable Go Game Images Recognition Under Strong Light Attack

Abstract

Go is a popular global game whose win or loss is only determined by the number of intersection points surrounded by black or white pieces. Among all the counting methods, the traditional manual counting method is time-consuming. Additionally, the current Go game images recognition technology cannot endure light reflection attacks or extreme image capture angles effectively. In this paper, a reliable Go game images recognition method is proposed which not only can resist light reflection attacks but also can endure various image capture angles. To obtain this goal, we propose a detection method based on the optimized CNNs (Convolutional Neural Network) framework. Experiments on recognizing 3220 images show that the average accuracy with our proposed method is over 99.99%, which is 22 times better than the accuracy of the stateof-the-art approach on Go game images recognition. Besides, our study provides potential references for the recognition of interfered small objects in groups that have few features. It provides a reference in similar application scenarios such as the detection of animal crowds, industrial parts, physiological tissues, and micro-particles.INDEX TERMS Go board detection, Go pieces recognition, CNN, model ensemble

Go-Game Image Recognition Based on Improved Pix2pix

“…Neto [8] proposed a method for identifying synthetically generated chess images in Blender using a Python API, achieving 97% accuracy in chess-piece classification by fine-tuning the VGG-16 convolutional network. Zhuo [9] proposed a Go-image-recognition method based on Yolo v5 that could resist the influence of light reflection, but the model was more complex and required higher computational performance. The above methods had the advantages of better model accuracy and robustness.…”

“…where TP is true positive, FP is false positive, FN is false negative, and TN is true negative. In addition, in order to pay more attention to the error rate of the chess-piece recognition, this study referred to [9] and added the following three evaluation indicators: AERIP, AERI, and AERTD. AERIP is the average error rate of erroneous intersection points, AERI is the average error rate of the images where errors appeared, and AERID is the average error rate of the images where errors could disrupt the judgment of the winner.…”