A Hierarchical Predictive Coding Model of Object Recognition in Natural Images

M W Spratling

doi:10.1007/s12559-016-9445-1

Abstract

Predictive coding has been proposed as a model of the hierarchical perceptual inference process performed in the cortex. However, results demonstrating that predictive coding is capable of performing the complex inference required to recognise objects in natural images have not previously been presented. This article proposes a hierarchical neural network based on predictive coding for performing visual object recognition. This network is applied to the tasks of categorising hand-written digits, identifying faces, and locating cars in images of street scenes. It is shown that image recognition can be performed with tolerance to position, illumination, size, partial occlusion, and within-category variation. The current results, therefore, provide the first practical demonstration that predictive coding (at least the particular implementation of predictive coding used here; the PC/BC-DIM algorithm) is capable of performing accurate visual object recognition.

Highlights

Localising and identifying items in visual scenes is of fundamental importance for many activities carried out by humans and other species
The current results show that a particular implementation of Predictive coding (PC) algorithm1 can locate cars in natural images of street scenes, identify individuals from their face, and can categorize numbers in images of hand-written digits
The following parameters were used: the similarity threshold for the clustering performed on the image patches was set equal to κ = 0.85; the threshold on the number of patches in each cluster was set equal to λ = 0; and the standard deviation of the Gaussian used to pre-process both the images and receptive fields (RFs) of the first processing-stage was set equal to σ = 4 pixels

Summary

Introduction

Localising and identifying items in visual scenes is of fundamental importance for many activities carried out by humans and other species. To solve this complex computational task, the brain is required to perform perceptual inference in order to find the most likely causes of the visual input. This process of object recognition is believed to be performed by a hierarchy of cortical regions along the ventral occipitotemporal pathway [1,2,3,4]. The current results show that a particular implementation of PC (the PC/BC-DIM) algorithm can locate cars in natural images of street scenes, identify individuals from their face, and can categorize numbers in images of hand-written digits

Methods

Results

Conclusion