The exponential growth of digital video content across educational, professional, and social platforms has created a significant challenge: users struggle to efficiently access and comprehend key information embedded within lengthy, information-rich videos. This issue is particularly acute for learners and professionals who often require rapid retrieval of salient concepts, visual explanations, or critical updates without viewing entire recordings. Existing summarization techniques predominantly rely on text extraction from audio transcripts, frequently overlooking essential visual elements such as slides, diagrams, and transitions that are vital for complete understanding. In response, this work presents a multimodal video summarization framework that integrates mathematical keyframe extraction with advanced generative AI models for text summarization. The system employs scene segmentation and content similarity analysis to identify and select representative keyframes-distinct video frames that encapsulate important visual information. These are synchronized with structured text summaries generated using state-of-the-art models, including BART (Bidirectional and Auto-Regressive Transformers), PEGASUS (Pre-training with Extracted Gap-Sentences for Abstractive Summarization), and Whisper (a multilingual automatic speech recognition model).