Back to Search
Start Over
Depth in convolutional neural networks solves scene segmentation
- Source :
- PLoS Computational Biology, PLoS Computational Biology, 16(7), PLoS Computational Biology, 16(7):e1008022. Public Library of Science, PLoS Computational Biology, Vol 16, Iss 7, p e1008022 (2020), PLOS Computational Biology
- Publication Year :
- 2019
-
Abstract
- Feed-forward deep convolutional neural networks (DCNNs) are, under specific conditions, matching and even surpassing human performance in object recognition in natural scenes. This performance suggests that the analysis of a loose collection of image features could support the recognition of natural object categories, without dedicated systems to solve specific visual subtasks. Research in humans however suggests that while feedforward activity may suffice for sparse scenes with isolated objects, additional visual operations ('routines') that aid the recognition process (e.g. segmentation or grouping) are needed for more complex scenes. Linking human visual processing to performance of DCNNs with increasing depth, we here explored if, how, and when object information is differentiated from the backgrounds they appear on. To this end, we controlled the information in both objects and backgrounds, as well as the relationship between them by adding noise, manipulating background congruence and systematically occluding parts of the image. Results indicate that with an increase in network depth, there is an increase in the distinction between object- and background information. For more shallow networks, results indicated a benefit of training on segmented objects. Overall, these results indicate that, de facto, scene segmentation can be performed by a network of sufficient depth. We conclude that the human brain could perform scene segmentation in the context of object identification without an explicit mechanism, by selecting or “binding” features that belong to the object and ignoring other features, in a manner similar to a very deep convolutional neural network.<br />Author summary To what extent do Deep Convolutional Neural Networks exhibit sensitivity to scene properties (e.g. object context) during object recognition, and how is this related to network depth? Linking human visual processing to performance of feed-forward DCNNs with increasing depth, our study explored if and how object information is differentiated from the backgrounds they appear on. We show that with an increase in network depth, there is a stronger selection of parts of the image that belong to the target object, compared to the rest of the image. In other words, network depth facilitates scene segmentation. Given that the operations of a very deep network can be performed by a recurrent network, we speculate that the human brain could perform scene segmentation, in the context of object identification, without an explicit mechanism using recurrent processing.
- Subjects :
- 0301 basic medicine
Male
genetic structures
Computer science
Vision
Social Sciences
Convolutional neural network
Visual processing
Feedforward Neural Networks
0302 clinical medicine
Cognition
Learning and Memory
Human Performance
Psychology
Segmentation
Biology (General)
Visual Cortex
Ecology
Artificial neural network
05 social sciences
Cognitive neuroscience of visual object recognition
Brain
Eukaryota
Signal Processing, Computer-Assisted
Semantics
Identification (information)
Computational Theory and Mathematics
Pattern Recognition, Visual
Feature (computer vision)
Modeling and Simulation
Vertebrates
Visual Perception
Feedforward neural network
Female
Sensory Perception
Research Article
Adult
Computer and Information Sciences
Adolescent
Neural Networks
QH301-705.5
ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION
Context (language use)
050105 experimental psychology
Birds
03 medical and health sciences
Cellular and Molecular Neuroscience
Young Adult
Memory
Genetics
Humans
Animals
Learning
0501 psychology and cognitive sciences
Molecular Biology
Ecology, Evolution, Behavior and Systematics
Behavior
business.industry
Cognitive Psychology
Organisms
Reproducibility of Results
Biology and Life Sciences
Pattern recognition
Recognition, Psychology
Linguistics
Object (computer science)
030104 developmental biology
Amniotes
Cognitive Science
Perception
Noise (video)
Artificial intelligence
Neural Networks, Computer
business
Visual Object Recognition
Zoology
030217 neurology & neurosurgery
Neuroscience
Subjects
Details
- ISSN :
- 15537358 and 1553734X
- Volume :
- 16
- Issue :
- 7
- Database :
- OpenAIRE
- Journal :
- PLoS computational biology
- Accession number :
- edsair.doi.dedup.....59b136d437993d718fc6b8fab0d7bbd8