Urban parks and tree-lined streets are major components of urban greenspaces, as well as the most frequently used public spaces for senior citizens. Studies have shown significant associations between urban greenspaces and various health outcomes of older adults. However, most of them focused on urban parks or overall vegetation coverage, and few have considered the impact of street greenery. The lack of research attention on the latter is partly because of no method objectively measuring greenery exposure on streets, especially from a pedestrian-centered perspective. In the current study, we recruited 1161 adults aged 60 or above who lived in 12 housing estates in Wuhan, China, and collected their socio-demographic data and 7-day physical activity data. Streetscape photos were taken by trained researchers on sidewalks of all streets in the 800-m buffers from these housing estates. The pedestrian-centered street greenery exposure was extracted from these photos with the machine learning technique of convolutional neural networks along with the pyramid pooling module. Multilevel logistic regression models were conducted to examine the association of the frequency (≥4 days vs. < 4 days) and total duration (≥300 mins vs. <300 mins) of physical activity with street greenery. Park area, population density, street connectivity, and land use mix within the buffer zone, as well as individual factors, were included as covariates in the models. Results showed that street greenery was positively associated with the odds of achieving 300 mins or more of physical activity per week, but the park area was not. Furthermore, street connectivity and land use mix were positively associated with both the frequency and total time of physical activity. Unexpectedly, population density was negatively associated with the frequency and total time of physical activity. Therefore, adding street greenery or improving existing street greenery can be a vital environment-intervention strategy to create an aging-friendly urban environment.