Exponential growth of multimedia on cyberspace and availability of sophisticated image editing tools have raised the propagation of forged images, leading to a major threat to information integrity and authenticity. Among the conventional image forgery techniques, copy-move forgery (CMF), involving copying and pasting a region within the same image to conceal or duplicate an object, remains the most challenging to detect due to its context-preserving nature and presence of the same noise pattern on the forged regions. Further, CMF is more challenging to detect in the presence of different post-processing attacks. To counter these challenges, we propose a deep learning-based framework, ReLU-EfficientNet, designed for robust CMF detection under post-processing attacks. The proposed method enhances the EfficientNetB0 architecture with the introduction of ReLU activation to enhance discriminative feature extraction, accelerate convergence, and mitigate vanishing gradient problems. Further, we also introduce three dense layers to refine learned features, resulting in better separability, improved generalization, and robust classification performance. We evaluated the performance on two diverse benchmark datasets, MICC-F2000 and CASIA v2.0. The accuracy of 97% on MICC-F2000 and 96% on CASIA v2.0 illustrates the competency of the proposed method over several contemporary CMF detection methods. Furthermore, our method exhibits minimal overfitting and stable learning, thus providing strong generalization. These findings highlight the effectiveness and computational efficiency of our ReLU-EfficientNet method for CMF detection.