Vision AI has made remarkable progress in recent years, especially in applications across a range of scenarios such as security surveillance and behavior analysis.
For the specific task of recognizing "a person smoking," vision AI can identify the person and the objects around them and infer whether they are smoking.
However, traditional vision AI can usually only locate objects precisely; it cannot understand their actions or behavior in detail.
To solve this problem, AIWin combined vision AI with data analytics to automatically search for the best model and parameter configuration,
improving the accuracy and efficiency of recognizing "smoking in progress." Throughout the process, the "smoking model" was trained without writing a single line of code. The logical judgment is left entirely to the AI.
This combination of technologies not only serves surveillance systems, but can also be applied to scenarios such as health monitoring in public places and the management of no-smoking areas, helping to create smarter,
more precise environmental monitoring systems.
*Image: a comparison of traditional AI versus traditional AI combined with data analytics.
Eight images in total; in fact, images 3 and 6 should not count as smoking in progress.