Estimating the axial force of rock bolts is vital for the stability and safety of underground excavations. Current methods suffer from being labor-intensive, time-consuming, and prone to inaccuracies. This paper presents a novel method for estimating the axial force of a rock bolt, a critical element in underground engineering, by analyzing the deformation field of the bearing plate. This innovative approach combines 3D laser scanning for data acquisition with a developed convolutional neural network (CNN) model for data processing, resulting in an efficient and non-destructive external monitoring technique. The proposed method was verified with 27 laboratory experiments. The results showed an accurate estimation of the axial forces of rock bolts with an average error of ± 5 kN, underlining the method's potential for in-situ application in underground engineering. This research not only contributes to the development of intelligent systems in rock mechanics and engineering but also has significant implications for industries, such as mining, tunneling, and underground engineering. Highlights: The study introduces a novel method that utilizes 3D laser scanning and a developed CNN model to accurately estimate the axial force of rock bolts. The proposed method achieves accurate estimations of rock bolts' axial forces with an average error of ± 5 kN. Future work includes the integration of object detection algorithms for large-scale 3D laser scanning and the creation of a database of various bearing plates to enhance the method's robustness. [ABSTRACT FROM AUTHOR]