The goal of image inpainting is to restore missing or damaged regions of an image by inferring plausible content from the surrounding context. In this work, we propose a deep learning-based image inpainting approach utilizing Partial Convolution to reconstruct occluded areas effectively. The core idea of this method is to distinguish between valid and missing pixels during the convolution process, ensuring that each output is computed based solely on valid (non-masked) inputs. This is achieved through a partial convolution operation, where the convolution is masked and re-normalized at every spatial location. A critical technical component of the approach is the automatic mask update mechanism, which updates the binary mask at each convolutional layer to track newly filled regions, allowing the network to progressively expand the valid area throughout the inpainting process. This dynamic updating makes the model robust to irregular and arbitrary hole shapes. We evaluated the proposed method on benchmark datasets using standard evaluation metrics including Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM). Experimental results demonstrate that our approach achieves superior performance compared to recent state-of-the-art inpainting methods, both quantitatively and qualitatively.