Abstract

Nowadays regular exercises are needed for preventing and maintaining the health from diseases. Analyze and give the best exercises based on the input to the user which then presents them the concatenated video. This paper integrates an LLM, in this case GPT-3.5 for NLP, which enables to interpret user inputs for the Weaviate vector database. The results were stitched video which is a combination of videos taken from the S3 bucket. The stitched video didn't result in any major bugs or quality issues. It is relevant to the user inputs such as preference, weaknesses and more. This application is focused to helps people doesn't know how and where to begin get a basic idea of exercises they can achieve their goals regarding their fitness. The videos selected and stitched together to make it easier for person viewing it to understand how the exercises work as well as do them correctly. The final output video is personalized to the user's preferences and skill level and so it will meet their objectives making the workout experience, better fitness, enjoyable and effective.