A model of amacrine cells for orientation detection

As the most studied sensory system, the visual system plays an important role in our understanding of brain functions. Biological researchers have divided the nerve cells in the retina into dozens of visual channels carrying various characteristics based on visual features. Although orientation-selective cells have been identified in the retinas of various animals, the specific neural circuits of such cells have been controversial. In this study, a new simple and efficient orientation detection model based on the perceptron is proposed to restore the neural circuitry of orientation-selective cells in the retina. The performance of this model is experimentally compared with that of the convolutional neural network for image orientation recognition, and the results verify that the proposed model offers very good orientation detection. The proposed perceptron-based orientation detection model provides a new perspective to explain the neural circuits of orientation-selective cells.