This dataset provides high-resolution dust distribution maps of the Galactic plane (\(|Z| < 25\) pc). It covers the full \(360^\circ\) range of Galactic longitude and reaches distances up to 7 kpc from the Sun. The dataset includes three dust density maps with different spatial resolutions: a Large Region (LR, 100 pc), a Medium Region (MR, 50 pc), and a Small Region (SR, 10 pc). These dust density distribution maps are constructed based on Red Clump stars selected from near-infrared (2MASS, VVV, UKIDSS) and mid-infrared (ALLWISE, GLIMPSE) photometry, combined with the Gaia BP/RP stellar sample from Zhang & Green (2025) and the Red Clump sample from Lucey et al. (2020). A U-Net convolutional neural network is employed to invert the line-of-sight extinction distributions of these samples into dust density distributions. This dataset is of significant scientific value for precise Galactic extinction correction, tracing spiral arm structures, and studying the interplay between the interstellar medium and star formation.