In sand casting, treating the mold surface with chemical coatings is essential to guarantee structural integrity prior to the introduction of molten metal. This activity has conventionally been performed manually, necessitating staff to apply chemical layers manually. Nonetheless, manual coating is labor-intensive and offers significant health risks to workers, including respiratory problems and skin burns from extended exposure to hazardous materials. This leads to inconsistent coating thickness and diminished efficiency, especially in extensive industrial activities. This research provides a deep learning framework that combines edge detection and depth estimation for the automatic identification of mold surfaces to overcome these constraints. The technology employs sophisticated computer vision methods to produce precise surface masks for directing autonomous painting units. A dataset obtained from PSG Foundry, Neelambur, was utilized to train and evaluate the model. Experimental findings indicate that the ensemble methodology attains elevated precision in identifying mold peripheries and coating areas, consequently diminishing reliance on manual labor. This technology provides substantial industrial advantages, such as greater worker safety, reduced chemical exposure, increased operating efficiency, and uniform coating quality. In addition to mold preparation, the technology facilitates extensive automation in foundries, with potential applications in robotic spraying and fault detection systems.