Mangrove forests play a vital role in coastal ecosystems by mitigating climate change, protecting shorelines, and supporting biodiversity. However, these critical habitats are declining globally due to encroachment, deforestation, urbanization, and aquaculture expansion. Regular and detailed monitoring is essential for effective conservation and restoration efforts. While freely available satellite imagery, such as from Sentinel-2, provides broad coverage, its coarse spatial resolution limits its usefulness for detailed mangrove monitoring. To address this gap, this study develops an unmanned aerial vehicle (UAV)-assisted super-resolution framework to reconstruct high-resolution historical imagery for monitoring mangrove restoration at Saphan Hin, Phuket, Thailand. We acquired high-resolution UAV imagery in 2024 to calculate precise vegetation indices and canopy cover metrics. These UAV-derived products are used to train a super-resolution deep learning algorithm. The trained model is then applied to upscale historical satellite imagery from Landsat and Sentinel-2 archives (2004–2024), creating an enhanced temporal series. From this reconstructed data, we extract key indicators of restoration trajectories, including canopy density and phenological patterns. This approach enables a direct, high-resolution comparison of ecosystem development between actively restored areas and zones of natural recovery. The primary goal of this study is to demonstrate a practical, scalable workflow that leverages AI to transform low-resolution satellite data into actionable, site-specific insights. By reconstructing past canopy dynamics in the absence of historical high-resolution data, this framework provides a powerful, non-invasive tool for long-term monitoring. It holds significant potential to improve the management of coastal restoration projects, enhance the accuracy of blue carbon accounting, and support the valuation of ecosystem services in vulnerable coastal regions.