In various enterprises, the security data are manually entered by multiple security workers at different times, which can lead to disordered data in the system, and also, analysis decision files need to be manually processed. To solve these problems, this paper presents a decision system for key location security information based on active perception and intelligent analysis. First, the designed system is developed based on C/S framework. Then, a deep learning perception model is constructed. Experimental results show that this system can significantly reduce the heavy workload of security inspectors, and reduce the frequency of safety accidents.