The rapid progress in deep learning has spurred significant advancements in the integration of computer vision and natural language processing, especially in the realm of image captioning applications. This paper introduces a novel approach that utilizes image captioning capabilities specifically for violence detection. While traditional methods for detecting violence in videos primarily depend on computer vision techniques, such as analyzing frames one by one, our method transforms video content into meaningful textual descriptions. This shift enables a new paradigm in video analysis and classification. Unlike conventional approaches that focus on processing visual features directly, our method focuses on understanding contextual information by utilizing image captioning for video frames and uses this information to detect violent scenes. We assess our approach using the Real-Life Violence Situations Dataset, demonstrating its effectiveness through accuracy measurements and confusion matrix analysis. The results suggest that our method presents a promising alternative to traditional violence detection techniques, potentially paving the way for new applications in security, content moderation, and public safety. This study not only advances the field of video analysis but also highlights the versatility of image captioning technologies in addressing real-world challenges.