<p>Full wavefield-based damage detection method has received widespread attention, but the applications in composite structures still face big challenges for the requirements of baseline and full wavefields with dense sensing points. In this paper, an intelligent framework of guided wave-based damage assessment method for composite laminates is proposed. First of all, a physics-oriented deep learning model of long short-term memory (LSTM) is trained to predict full wavefield from incomplete measurements without requiring material parameters. Afterward, based on the smart sensing method, an intelligent damage assessment method is proposed to reconstruct a pseudo-baseline and visualize the damaged region. Numerical simulations and experimental tests have been conducted to investigate the effectivity of the proposed smart sensing method, which can predict the full wavefield beyond the sensing region with high Pearson correlation coefficients (&gt; 0.98) and reconstruct a pseudo-baseline wavefield for damage detection. Results show the proposed damage detection method can visualize the single/multiple damaged regions in composite laminates with precise locations and sizes. In summary, this study provides an essential and promising approach for damage assessment by sensing and analyzing the guided waves in composite laminates.</p>