Corpus ID: 218596308

RetinotopicNet: An Iterative Attention Mechanism Using Local Descriptors with Global Context

  title={RetinotopicNet: An Iterative Attention Mechanism Using Local Descriptors with Global Context},
  author={Thomas Kurbiel and Shahrzad Khaleghian},
Convolutional Neural Networks (CNNs) were the driving force behind many advancements in Computer Vision research in recent years. This progress has spawned many practical applications and we see an increased need to efficiently move CNNs to embedded systems today. However traditional CNNs lack the property of scale and rotation invariance: two of the most frequently encountered transformations in natural images. As a consequence CNNs have to learn different features for same objects at… Expand
1 Citations
Study of Convolutional Neural Networks for Global Parametric Motion Estimation on Log-Polar Imagery
Experiments on existing image datasets using synthetic image deformations reveal that standard CNNs can successfully learn to estimate global parametric motion on log-polar images with accuracies comparable to or better than with Cartesian images. Expand


Beyond Cartesian Representations for Local Descriptors
This work proposes to extract the “support region” directly with a log-polar sampling scheme and shows that this provides a better representation by simultaneously oversampling the immediate neighbourhood of the point and undersampling regions far away from it, and is particularly amenable to learning descriptors with deep networks. Expand
Learning Multiple Layers of Features from Tiny Images
It is shown how to train a multi-layer generative model that learns to extract meaningful features which resemble those found in the human visual cortex, using a novel parallelization algorithm to distribute the work among multiple machines connected on a network. Expand
Polar Transformer Networks
PTN combines ideas from the Spatial Transformer Network (STN) and canonical coordinate representations and is a network invariant to translation and equivariant to both rotation and scale, which is extensible to 3D which is demonstrated through the Cylindrical Trans transformer Network. Expand
SaccadeNet: A Fast and Accurate Object Detector
Among all the real-time object detectors, the proposed SaccadeNet achieves the best detection performance, which demonstrates the effectiveness of the proposed detection mechanism. Expand
Effnet: An Efficient Structure for Convolutional Neural Networks
EffNet is a novel convolution block which significantly reduces the computational burden while surpassing the current state-of-the-art and is created to tackle issues in existing models such as MobileNet and ShuffleNet. Expand
A review of log-polar imaging for visual perception in robotics
This paper surveys the application of log-polar imaging in robotic vision, particularly in visual attention, target tracking, egomotion estimation, and 3D perception and to help readers identify promising research directions. Expand
MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
This work introduces two simple global hyper-parameters that efficiently trade off between latency and accuracy and demonstrates the effectiveness of MobileNets across a wide range of applications and use cases including object detection, finegrain classification, face attributes and large scale geo-localization. Expand
Effective Approaches to Attention-based Neural Machine Translation
A global approach which always attends to all source words and a local one that only looks at a subset of source words at a time are examined, demonstrating the effectiveness of both approaches on the WMT translation tasks between English and German in both directions. Expand
Robust image registration using log-polar transform
  • G. Wolberg, Siavash Zokai
  • Mathematics, Computer Science
  • Proceedings 2000 International Conference on Image Processing (Cat. No.00CH37101)
  • 2000
The algorithm estimates the affine transformation parameters necessary to register any two digital images misaligned due to rotation, scale, shear, and translation using a variation of the Levenberg-Marquadt nonlinear least squares optimization method, which yields a robust solution that precisely registers images with subpixel accuracy. Expand
Turning the world around: Patterns in saccade direction vary with picture orientation
The findings show that a horizon bias is robust and affected by both the distribution of features and more global representations of the scene layout, though this pattern varies in interiors and during recognition of previously seen pictures. Expand