Camera?based observation of forest canopies allows for low?cost, continuous, high temporal?spatial resolutions of plant phenology and seasonality of functional traits. In this study, we extracted canopy color index (green chromatic coordinate, Gcc) from the time?series canopy images provided by a digital camera in a deciduous forest in Massachusetts, USA. We also measured leaf?level photosynthetic activities and leaf area index (LAI) development in the field during the growing season, and corresponding leaf chlorophyll concentrations in the laboratory. We used the Bayesian change point (BCP) approach to analyze Gcc. Our results showed that (1) the date of starting decline of LAI (DOY 263), defined as the start of senescence, could be mathematically identified from the autumn Gcc pattern by analyzing change points of the Gcc curve, and Gcc is highly correlated with LAI after the first change point when LAI was decreasing (R2 = 0.88, LAI < 2.5 m2/m2); (2) the second change point of Gcc (DOY 289) started a more rapid decline of Gcc when chlorophyll concentration and photosynthesis rates were relatively low (13.4 ± 10.0% and 23.7 ± 13.4% of their maximum values, respectively) and continuously reducing; and (3) the third change point of Gcc (DOY 295) marked the end of growing season, defined by the termination of photosynthetic activities, two weeks earlier than the end of Gcc curve decline. Our results suggested that with the change point analysis, camera?based phenology observation can effectively quantify the dynamic pattern of the start of senescence (with declining LAI) and the end of senescence (when photosynthetic activities terminated) in the deciduous forest.