<p>Mobile edge computing (MEC) is expected as a promising technology to increase the computational capability of security monitoring internet of things (IoT) system which integrates various artificial intelligence (AI)-based applications such as intelligent video analysis, dynamic target tracking, etc. However, due to the complex wireless environment and dynamic computing demands, MEC may suffer inaccurate task offloading issue and result in performance degradation of security monitoring IoT system. In this paper, we firstly design a cooperative MEC framework for IoT devices and unmanned monitoring vehicles (UMVs) collaboratively to process tasks with different priority levels. Then, we propose a priority-based task offloading method to maximize the system utility under energy constraints of IoT devices and UMVs. Given that the policy of task offloading depends on the system state, we formulate the task offloading problem as a Markov decision process (MDP) and propose an adaptive long-term optimization algorithm to solve it. Such algorithm combines deep deterministic policy gradient (DDPG) and a reward estimation method in case the reward function may be lost due to communication errors. Finally, numerical results validate the effectiveness and superiority of our proposed scheme compared to traditional algorithms.</p>