The proliferation of digital video content in today's digital age has given rise to an increasing concern regarding the authenticity and integrity of video data. Video forgery, such as splicing, tampering, or manipulation, poses a significant threat to the credibility of multimedia content. This paper introduces a method for detecting spatial-video forgeries by leveraging a customized ResNet50 deep CNN. Within this investigation, we address the challenge of detecting spatial video forgeries by leveraging the power of deep learning We propose a Modified ResNet50 architecture that has been fine-tuned for the specific task of video forgery detection. The modified architecture is trained on a varied dataset comprising genuine and manipulated video frames, allowing it to learn intricate features and patterns indicative of forgery. Our approach benefits from the ResNet50 deep architecture, This approach naturally grasps hierarchical and abstract features within video frames. The results obtained from experiments on benchmark datasets validate the efficacy of our method, illustrating its precision in identifying spatial video forgeries. We evaluate the effectiveness of our model by comparing its performance with cutting-edge methods, showcasing its superior accuracy and robustness. Moreover, we evaluate the model's performance under various forgery scenarios, including copy-move, object insertion, and object removal.