Feedforward object-vision models only tolerate small image variations compared to human.
Masoud GhodratiAmirhossein FarzmahdiKarim RajaeiReza EbrahimpourSeyed-Mahdi Khaligh-RazaviPublished in: Frontiers Comput. Neurosci. (2014)
Keyphrases
- feed forward
- object models
- neural nets
- input image
- three dimensional objects
- image retrieval
- image regions
- image features
- keypoints
- image segmentation
- artificial neural networks
- back propagation
- multiscale
- image content
- visual input
- spatial relations
- single image
- image classification
- lighting conditions
- object parts
- vision system
- real time
- partial occlusion
- multiple objects
- bounding box
- d objects
- image representation
- global shape
- pixel level
- ground plane
- computational models
- feature points
- neural architecture
- adaptive neural
- neural network model
- artificial intelligence