This invention discloses a method and
system for identifying abnormal behaviors in urban social environments based on
image analysis, belonging to the field of
image analysis technology. It includes the following steps:
data acquisition, obtaining environmental
weather data for weekdays, holidays, different lighting times (
morning,
noon,
evening), and weather conditions (sunny, rainy, foggy), as well as event
feature data from school building corridors,
community squares, and entrances / exits of workplaces, to obtain
event data. This invention, through the setting of a multi-dimensional
environmental data acquisition module, first performs multi-dimensional
data acquisition, combining real-world and simulated scenarios, and finally inputs the image to be detected. The model outputs the
event type, target location, and feature attributes, determines anomalies according to preset rules, associates keyframes for storage and triggers warnings, and outputs the recognition results. This
system ensures that the model can adapt to complex spatiotemporal and environmental changes in the real world, avoids
overfitting caused by training in a single
scenario, and significantly improves generalization ability.