A novel method for splicing video forgery detection in the digital domain is proposed. Our method combines the ResNext architecture with region_based segmentation to resolve the issues with multimedia data authenticity. After preprocessing, which includes resizing and normalization, video frames are segmented hierarchically using Spatial_Constraint Fuzzy C_Means (SCFCM). Then, for features extraction and classification, the ResNext deep learning model is used. Using deep learning capabilities and structural segmentation, this integrated methodology presents a standard solution for robust splicing video forgery detection. Its effectiveness in addressing digital video forensic issues is confirmed by experimental evaluations on benchmark datasets. Superior accuracy and strength in forgery identification are highlighted by thorough comparisons with state-of-the-art techniques, and results show precision in detecting spatial video forgeries.