State of Health (SOH) estimation of lithium-ion batteries is crucial for ensuring operational safety and reliability. Traditional SOH estimation methods often rely on long-term cycling or operating data, which can be time-consuming and limited in accuracy. This study proposes a novel approach that combines Electrochemical Impedance Spectroscopy (EIS) features with a Convolutional Neural Network (CNN) to estimate battery SOH. EIS data were collected across various degradation states, and both magnitude and phase spectra were converted into image-like representations to serve as inputs for the CNN model. Leveraging the CNN’s ability to automatically extract and learn complex patterns, the proposed method achieves accurate SOH predictions without requiring prior knowledge of battery history or internal parameters. Experimental results demonstrate that this approach enables fast, non-invasive, and data-driven SOH estimation, offering strong potential for real-time battery health monitoring applications.