Digital image manipulation has become increasingly sophisticated, leading to a rise in image forgery cases. In this work proposed model to detect splice and copy-move manipulations using deep learning model, Enhanced DenseNet201 and VGG19, on the widely used CASIA 2.0 and CASIA 1.0 dataset. The proposed methodology focuses on enhancing the accuracy of forge image classification through the utilization of advanced convolutional neural networks (CNNs). Enhanced DenseNet201 and VGG19 models have fine-tuned and trained on the CASIA 1.0 and CASIA 2.0 dataset, which comprises authentic and tampered photos. The models are equipped to identify instances of splice and copy-move forgeries. By leveraging the deep features extracted by these models, the proposed approach demonstrates remarkable performance in accurately distinguishing between genuine and manipulated images. The evaluation of the models is carried out using standard evaluation metrics accuracy, precision, recall, and auc. The achieved results underscore the effectiveness of the Enhanced DenseNet201 approach, with accuracy, precision, recall, and auc values of 0.9181, 0.9157, 0.9113, and 0.9357, respectively. These metrics showcase the models’ robustness and capability to classify splice and original images with high precision and recall.