This paper proposes a method of allocation of cognitive resource for efficient real-time multiple object recognition in robot vision. For human beings and even high-performance computers, the number of recognizable objects is limited at an instant time because cognitive and computational resources are limited. Thus, real-time reallocation of computational resources to the multiple objects is logical for robot to recognize them in real time under dynamically changing environment. We propose such dynamical reallocation scheme for robots recognize multiple objects effectively in real time. Positions of object models are represented by the genes of GA. The individuals of GA are allocated for recognizing multiple objects depending on the situation and priority of the object recognition. The proposed method is able to improve the speed of multiple object recognition. Some simulations of multiple object recognition will show the effectiveness of the method.