Model of Human Visual Cortex Inspired Computational Models for Visual Recognition

In this paper, we are mostly interested in investigating how the study and discovery of the human visual cortex could be utilised to improve the computational models for visual recognition by computer vision. Many of the brain perceptual abilities in vision have corresponding algorithms exist in computer vision, and in this paper we discuss three such models. First we present a model that has the ability for iterative bottom-up/top-down recognition, and experimental results on applying the model for facial landmark detection has shown improved accuracy over benchmark approaches. Second we introduce a new SOM model that could be deep and invariant, which could achieve significantly improved digit recognition accuracy over traditional SOM. And third we show how the convolutional neural network could be combined with linear coding based architecture, where experimental results show that the proposed model could outperform many existing algorithms for image classification.