<p>Inpainting is a technique used to modify the visual components of photographs. It involves removing a specific part of an image and replacing it with new information that reduces any visible inconsistencies. This subject is of great importance; when used maliciously, inpainting can alter images undetectably, affecting areas like legal proceedings, news authenticity, and personal privacy. Hence, this paper addresses the challenge of detecting image inpainting, where a robust end-to-end deep learning framework for this purpose is presented. The proposed approach leverages advanced feature extraction techniques to enhance the model’s effectiveness, in addition to utilizing feature pyramid and a residual network backbone, that are tailored for image inpainting detection. The use of feature pyramid is highly beneficial for reducing intra-class differences by integrating features at multi-scale levels. In addition, a custom public dataset for detecting inpainting, consisting of (12k) images, was created to serve as a foundation for such work. The results obtained from assessments, with a detection accuracy of over (&gt;90%), on the recently created benchmark dataset highlight the strength and reliability of the proposed deep learning approach. These findings demonstrate the model’s strong ability to accurately detect picture manipulation and represent a significant improvement in the tools accessible to analysts who are concerned with protecting the authenticity of digital images.</p>