The experiment of measuring magnelie fields using three-dimensional Helmholtz coils is a eritiealeleetromagnetism exercise in university physies currieula.lt aims to equip sudents with weak field measurementtechniques, verify the magnetie field superposition principle, and map the magnetie flux density distribution.This study integrates deep learning algorithms into the three-dimensional Helmholtz coil magnetie fieldmeasurement system. Under data-intensive experimental conditions,this innovation not only enhancesoperational efficieney but also demonstrates potential for non-destruetive testing of complex surface cracks onmetal sheets. As a pioneering case study, it offers novel interdiseiplinary approaches for merging universityphysics experiments with artificial intelligence ( Al) and academic competitions, while providing robust supportfor intelligent pedagogical reforms in physics laboratories.This work exemplifies the feasibility of integrating Alinto conventional teaching frameworks and foreshadows the intelligent evolution of future physics experimenteducation.