Exploring Daily Habits: Visual Life Logging to Decode Lifestyle Health

Living a healthy life is crucial for overall well-being and happiness, impacting quality of life and preventing diseases. Currently, lifelogging and egocentric datasets do not cater to lifestyle analysis, leaving a gap in this area of computer vision research. This study tackles the challenge of lifestyle analysis by creating a Visual Lifelogging Dataset for Lifestyle Analysis (VLDLA). The VLDLA includes images snapped every three seconds from 8:00 AM to 6:00 PM over seven days. Unlike existing datasets, our dataset is designed for lifestyle analysis by capturing activities in short intervals and covering the entire day.

Activities in each image are identified, and three time-changing factors, or latent fluents, associated with these activities measure the healthiness of the user's lifestyle. Scores for these factors are based on recognized activities, and the overall healthiness of the lifestyle for the day is determined by these scores. Testing showed that this method can effectively assess how healthy a user's lifestyle is.

Actions