Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Bioinspired CNNs for border completion in occluded images

Catarina P. Coutinho, Aneeqa Merhab, Janko Petkovic, Ferdinando Zanchetta, Rita Fioresi

TL;DR

The authors' CNN architecture, BorderNet, is evaluated on three occluded datasets (MNIST, Fashion-MNIST, and EMNIST) under two types of occlusions: stripes and grids.

Abstract

We exploit the mathematical modeling of the border completion problem in the visual cortex to design convolutional neural network (CNN) filters that enhance robustness to image occlusions. We evaluate our CNN architecture, BorderNet, on three occluded datasets (MNIST, Fashion-MNIST, and EMNIST) under two types of occlusions: stripes and grids. In all cases, BorderNet demonstrates improved performance, with gains varying depending on the severity of the occlusions and the dataset.

Bioinspired CNNs for border completion in occluded images

TL;DR

The authors' CNN architecture, BorderNet, is evaluated on three occluded datasets (MNIST, Fashion-MNIST, and EMNIST) under two types of occlusions: stripes and grids.

Abstract

We exploit the mathematical modeling of the border completion problem in the visual cortex to design convolutional neural network (CNN) filters that enhance robustness to image occlusions. We evaluate our CNN architecture, BorderNet, on three occluded datasets (MNIST, Fashion-MNIST, and EMNIST) under two types of occlusions: stripes and grids. In all cases, BorderNet demonstrates improved performance, with gains varying depending on the severity of the occlusions and the dataset.
Paper Structure (7 sections, 12 equations, 26 figures, 2 tables)

This paper contains 7 sections, 12 equations, 26 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Mathematical Modeling for border completion in V1
  3. Experiments
  4. Results
  5. Conclusions

Figures (26)

  • Figure 1: Association Fields
  • Figure 2: Numerical solutions of the geodesic equations for some values of $\gamma$
  • Figure 3: Implemented pipeline with Vanilla LeNet5 and BorderNet trained on unoccluded images of the three different datasets (MNIST/FashionMNIST/EMNIST) and tested on two types of occluded images (stripes/grid).
  • Figure 4: BorderNet oriented filters.
  • Figure 5: Sample of images occluded with stripes (top row) and grids (bottom row) used in the testing phase occluded.
  • ...and 21 more figures

Theorems & Definitions (1)

  • definition 1