Video summarization using deep learning

In today’s digital age, video content is rapidly increasing across platforms, making it challenging to consume and analyze lengthy videos efficiently. This project presents a Video Summarisation system using Deep Learning techniques to address this challenge. The system allows users to either upload a video or provide a URL link. For uploaded videos, the audio is extracted and transcribed using the Wav2Vec2ForCTC model combined with Wav2Vec2Processor, enabling accurate speech-to-text conversion. For YouTube links, the YouTube Transcript API is utilized to directly fetch the textual transcript. The resulting transcription undergoes summarization using a BERT-based algorithm, providing a condensed, meaningful summary. Furthermore, the summarized content is translated into Kannada and other Indian languages using Google Translate API, making it accessible to a wider audience. The system is developed with a front-end based on HTML, CSS, and JavaScript, and a Python-powered backend. This tool offers an efficient, multilingual, and user-friendly solution for fast video content consumption.  

Keywords: Video Summarization, Deep Learning, Wav2Vec2ForCTC, BERT, YouTube Transcript API, Speech Recognition, Natural Language Processing, Summarization, Multilingual Translation, Google Translator.