Manuscript received March 14, 2025; revised April 8, 2025; accepted June 9, 2025; published September 17, 2025.

Abstract—Image processing methods are now widely used in the world in civilian fields such as handwriting recognition, face recognition, gender classification, self-driving cars, disease diagnosis, etc. The paper introduces a target reporting method using machine learning algorithms and image processing methods to identify bullet marks on targets, supporting shooting training. Bullet mark images taken by camera will be digitized and the K-Nearest Neighbor machine learning classification algorithm will be applied. The algorithm models after being trained give quite accurate results in identifying bullet mark locations in the target mark circle, thereby calculating the total score achieved by the shooter after shooting, contributing to improving shooting training capacity. The overall average accuracy rate achieved is over 98.5%, which is comparable to other techniques such as Histogram of Oriented Gradients + Support Vector Machines (HOG + SVM) and Faster Recurrent Convolutional Neural Network (Faster R-CNN). However, the processing time when applying the K-Nearest Neighbor machine learning classification algorithm is much faster, meeting the requirements for fast processing in actual shooting training.

Keywords—machine learning classification, image processing, target score reporting

Cite: Minh H. Nguyen, Long Q. Trinh, and Tuan D. Nguyen, "An Approach to the Problem of Calculating Target Scores Using Camera to Support Shooting Training," Journal of Image and Graphics, Vol. 13, No. 5, pp. 502-514, 2025.

Copyright © 2025 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC-BY-4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.