Flowering phenology is a fundamental basis for landscape configuration and the maintenance of ornamental effects in garden plants, but traditional manual observation cannot efficiently acquire flowering dynamic information of multiple cultivars and multi-temporal phases. In this study, flowering time-series data of 40 garden chrysanthemum cultivars were extracted based on multi-temporal unmanned aerial vehicle red-green-blue (RGB) images, flowering time-series curves of garden chrysanthemums were constructed by using the piecewise cubic Hermite interpolating polynomial (PCHIP) method, and the flowering duration characteristics of each cultivar and population flowering configuration patterns were analyzed. The results show that the mean errors between the flowering periods identification of unmanned aerial vehicle and manual monitoring are within 2 d, which can efficiently replace the traditional discrete manual monitoring. The 40 garden chrysanthemum cultivars exhibit evident differences in flowering dynamics, and can be classified into four types, namely persistent type, rapid-bloom and rapid-decline type, slow-bloom type, and slow-decline type, containing 28, 5, 2, and 5 cultivars, respectively; cultivar combinations are classified into three types, namely highoverlap type, intermediate type, and staggered type, containing 426, 209, and 145 pairs, respectively. Based on three scenarios of festival display, routine greening, and tourist sightseeing, flowering configuration modes of festival type, routine type, and recreational type are constructed, respectively.