Biologically inspired visual computing: the state of the art
- PDF / 822,016 Bytes
- 15 Pages / 612.284 x 802.205 pts Page_size
- 48 Downloads / 269 Views
Biologically inspired visual computing: the state of the art Wangli HAO1,4, Ian Max ANDOLINA6, Wei WANG3,4,5,6, Zhaoxiang ZHANG
1,2,3,4
1
3
Research Center for Research on Intelligent Perception and Computing, Beijing 100190, China 2 National Laboratory of Pattern Recognition, CASIA, Beijing 100190, China CAS Center for Excellence in Brain Science and Intelligence Technology, CAS, Beijing 100190, China 4 University of Chinese Academy of Sciences, Beijing 100190, China 5 State Key Laboratory of Neuroscience, Shanghai 200031, China 6 Institute of Neuroscience, Chinese Academy of Sciences, Shanghai 200031, China
c Higher Education Press 2020
Abstract Visual information is highly advantageous for the evolutionary success of almost all animals. This information is likewise critical for many computing tasks, and visual computing has achieved tremendous successes in numerous applications over the last 60 years or so. In that time, the development of visual computing has moved forwards with inspiration from biological mechanisms many times. In particular, deep neural networks were inspired by the hierarchical processing mechanisms that exist in the visual cortex of primate brains (including ours), and have achieved huge breakthroughs in many domainspecific visual tasks. In order to better understand biologically inspired visual computing, we will present a survey of the current work, and hope to offer some new avenues for rethinking visual computing and designing novel neural network architectures. Keywords brain-inspired, vision, neural models, intelligence, novel neural networks
1
Consequently, designing new visual computing models inspired from the current perspective of visual cognition remains an urgent and challenging task. Historically, the development of visual computing has been heavily inspired by biological mechanisms. For example, local feature descriptors were first designed and inspired by the characteristics of the receptive fields (RFs) of the visual cortex, and achieved promising performance in many computer vision tasks. From the seminal Neocognitron [21], to HMAX [22], and finally deep learning, all of these models take inspiration from the hierarchical feature layering and processing mechanisms observed in biological visual systems. Although there are far too many research papers in each area to provide comprehensive treatments, we will review example studies in biologically inspired visual computing methods across four domains: • Biological neuron inspired visual computing. • Biological neural-circuit inspired visual computing. • Biological functional and cognition inspired visual computing. • Biological learning inspired visual computing.
Introduction
Computer vision is a discipline that studies how to make a machine “see”. It encompasses the use of cameras or other sensing devices and computers instead of biological eyes to perform classification, tracking, and measurement of objects. The main task of computer vision is to process captured pictures or videos and obtain actionable informati
Data Loading...
