With the rapid growth of video content, there is a growing need for smart ways to pick out the most useful frames for image processing tasks. This paper introduces a simple yet effective method for automatically extracting and grading frames from a video. The process starts by capturing frames at a low frame rate, either 1 frame per second (fps) or 0.5 fps, based on the user's choice. Each frame is then evaluated in two ways: First, we check how sharp or clear the image is using the Variance of Laplacian (VoL) method. Second, we use MediaPipe Face Mesh to see how well any faces in the frame are visible. These two scores-image clarity and face visibility-are combined using a custom ranking algorithm. Frames that are sharper and show clearer faces are ranked higher. Finally, the frames are saved in a specific folder, sorted from highest to lowest rank, giving you a clean and organized set of frames ready for further image processing work. © 2025 IEEE.
Язык оригиналаАнглийский
DOI
СостояниеОпубликовано - 2025
Событие5th International Conference on Power, Instrumentation, Control and Computing -
Продолжительность: 9 окт 202511 окт 2025
Номер конференции: 218525

конференция

конференция5th International Conference on Power, Instrumentation, Control and Computing
Сокращенное названиеPICC 2025
Период9/10/2511/10/25

ID: 150944456