Enhancing Video Capture in Mobile Applications for Power Saving Through Machine Learning

In today's digital age, people have become accustomed to using camera applications to record and share their life experiences through videos. Enhancing the user experience of mobile devices involves a critical aspect to minimize the power consumption associated with recording videos. This paper delves into the realm of processing and displaying power-efficient videos, all recorded in real-time through camera applications. Based on pixel-scaling methods, we introduce a frame-ratio predictor, utilizing machine-learning techniques to predict frame ratios efficiently. The results, as demonstrated on a commercial smartphone using four real-world videos, are remarkably promising.