Teacher-directed learning in view-independent face detection with mixture of experts using Cortex-Like Features

摘要

In this paper we propose view-independent face detection model using cortex-like features and classify patterns with Mixture of Expert, ME. Feature extraction of the model uses C1 Standard Model Feature, C1 SMF, motivated from biology and eigenface for dimensional reduction on the CMU PIE dataset. Since C1 Standard Model Feature and the combining method based on ME with teacher-directed learning for view invariance were jointly used and also has a biological motivation, the proposed model get high detection rate 97% at different face view images.