This system applies deep learning and Vision Transformer technology to examine gameplay video clips, specifically focusing on extracting and analyzing killshot moments. Users can train models locally on their own footage using labeled data (cheating vs. legitimate) and then run inference to receive confidence scores ranging from 0.0 (likely legitimate) to 1.0 (likely cheating). Frame-by-frame analysis is provided with results automatically persisted for later review.