| 摘要 |
The present invention is configured from: a recognition processing unit having a plurality of recognition processing modes; an evaluation unit for evaluating the performance of each process in the plurality of recognition processes possessed by the recognition processing unit; a statistical analysis unit for statistically analyzing a plurality of past evaluation results obtained from the evaluation unit, and determining, for each camera or each time slot, which recognition process is optimal; and a selection control unit characterized in determining, for an image obtained from the camera and in accordance with the result obtained by the statistical analysis unit, one or a plurality of re cognition processes from among the plurality of recognition processes possessed by the recognition processing unit, and causing recognition processing to be performed using the determined mode by the recognition processing unit. This enables a recognition mode optimal for an environment in which the camera is installed to be automatically set from past statistical results, and makes it possible to realize a reduction in manual adjustment costs and an improvement in recognition accuracy. |
