Bioinspired CNNs for border completion in occluded images

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

cs.CVMar 11, 2026

Bioinspired CNNs for border completion in occluded images

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

Organizations: Department of Pharmacy and Biotechnology, University of Bologna, Via San Donato 15, 40127 Bologna, Italy · Department of Mathematics, University of Ferrara, Via Ariosto 35, 44122 Ferrara, Italy · Numerical Analysis and Applied Mathematics, VIB - KU Leuven, ON5/B Herestraat 49 - Bus 4039, 3000 Leuven, Belgium

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.

Figures & tables

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).

Panel of Figure 5 — Figure 5: Sample of images occluded with stripes (top row) and grids (bottom row) used in the testing phase occluded.

	MNIST			FashionMNIST			EMNIST
(w,s)	LeNet5	BorderNet	Imp.*	LeNet5	BorderNet	Imp.*	LeNet5	BorderNet	Imp.*
(1,1)	0.974	0.974	99.90%	0.848	0.853	100.60%	0.989	0.985	99.70%
(1,10)	0.982	0.979	99.70%	0.854	0.867	101.50%	0.991	0.988	99.70%
(2,4)	0.939	0.960	101.80%	0.822	0.774	95.20%	0.962	0.976	101.30%
(3,3)	0.788	0.839	105.90%	0.689	0.683	100.70%	0.843	0.891	105.20%
(4,1)	0.573	0.630	108.90%	0.514	0.523	101.20%	0.673	0.697	103.70%

Table 1: Mean Accuracies of Vanilla LeNet5 and BorderNet for some examples of stripe occlusions for MNIST/FashionMNIST/EMNIST for different combinations of stripe width and stripe spacing (w,s) . Improvement (Imp.*) of BorderNet over Vanilla LeNet5 are the bootstrapped medians in percentage.

	MNIST			FashionMNIST			EMNIST
(w,s)	LeNet5	BorderNet	Imp.*	LeNet5	BorderNet	Imp.*	LeNet5	BorderNet	Imp.*
(1,1)	0.882	0.859	95.50%	0.722	0.717	99.60%	0.939	0.890	94.10%
(1,10)	0.966	0.973	106.60%	0.823	0.852	103.40%	0.979	0.984	100.50%
(2,4)	0.674	0.764	112.70%	0.549	0.675	123.30%	0.647	0.811	125.80%
(3,3)	0.384	0.412	107.50%	0.381	0.433	113.40%	0.398	0.413	102.00%
(4,1)	0.153	0.234	152.40%	0.174	0.304	177.50%	0.143	0.168	120.20%

Table 2: Mean Accuracies of Vanilla LeNet5 and BorderNet for some examples of grid occlusions for MNIST/FashionMNIST/EMNIST for different combinations of stripe width and stripe spacing (w,s) . Improvement (Imp.*) of BorderNet over Vanilla LeNet5 are the bootstrapped medians in percentage.

Panel of Figure 6 — Figure 6: Accuracy improvement of BorderNet with respect to Vanilla LeNet5 for occluded MNIST images with stripes (a) and grid (b).

Panel of Figure 7 — Figure 7: Accuracy improvement of BorderNet with respect to Vanilla LeNet5 for occluded FashionMNIST images with stripes (a) and grid (b).

Panel of Figure 8 — Figure 8: Accuracy improvement of BorderNet with respect to Vanilla LeNet5 for occluded EMNIST images with stripes (a) and grid (b).

Appendix figures & tables6 assets

Supplementary material from the paper’s appendix.

Appendix

Panel of Figure 9 — Figure 9: Accuracies with MNIST occlusion stripe spacing and width for the Vanilla LeNet5 and BorderNet.

Panel of Figure 10 — Figure 10: Accuracies with MNIST occlusion grid spacing and width for the Vanilla LeNet5 and BorderNet.

Panel of Figure 11 — Figure 11: Accuracies with FashionMNIST occlusion stripe spacing and width for the Vanilla LeNet5 and BorderNet.

Panel of Figure 12 — Figure 12: Accuracies with FashionMNIST occlusion grid spacing and width for the Vanilla LeNet5 and BorderNet.

Panel of Figure 13 — Figure 13: Accuracies with EMNIST occlusion stripe spacing and width for the Vanilla LeNet5 and BorderNet.

Panel of Figure 14 — Figure 14: Accuracies with EMNIST occlusion grid spacing and width for the Vanilla LeNet5 and BorderNet.

Explore similar work

OASIC: Occlusion-Agnostic and Severity-Informed Classification
Apr 5, 2026Kay Gijzen, Gertjan J. Burghouts, Daniël M. PeltImage Classification Occlusion Reasoning
Leiden University · TNO
Robust Onion: Peeling Open Vocab Object Detectors Under Noise
Jun 25, 2026Priyank Pathak, Mukilan Karuppasamy, Aaditya Baranwal +2Open-Vocabulary Object Detection VLM Robustness
UCF Institute of Artificial Intelligence, University of Central Florida (UCF)
Some Complexity Results for Robustness Verification for Binarized Neural Networks
Jun 17, 2026Harshit Goyal, Sudakshina DuttaNeural Network Verification Binary Neural Networks
Indian Institute of Technology Goa

Apr 5, 2026J/K move · Enter open · S save

OASIC: Occlusion-Agnostic and Severity-Informed Classification

Kay Gijzen, Gertjan J. Burghouts, Daniël M. Pelt

Leiden University · TNO

Image Classification Occlusion Reasoning

Open paper AskPDF arXiv

Severe occlusions of objects pose a major challenge for computer vision. We show that two root causes are (1) the loss of visible information and (2) the distracting patterns caused by the occluders. Our approach addresses both causes at the same time. First, the distracting patterns are removed at test-time, via masking of the occluding patterns. This masking is independent of the type of occlusion, by handling the occlusion through the lens of visual anomalies w.r.t. the object of interest. Second, to deal with less visual details, we follow standard practice by masking random parts of the object during training, for various degrees of occlusions. We discover that (a) it is possible to estimate the degree of the occlusion (i.e. severity) at test-time, and (b) that a model optimized for a specific degree of occlusion also performs best on a similar degree during test-time. Combining these two insights brings us to a severity-informed classification model called OASIC: Occlusion Agnostic Severity Informed Classification. We estimate the severity of occlusion for a test image, mask the occluder, and select the model that is optimized for the degree of occlusion. This strategy performs better than any single model optimized for any smaller or broader range of occlusion severities. Experiments show that combining gray masking with adaptive model selection improves

\text{AUC}_\text{occ}

by +18.5 over standard training on occluded images and +23.7 over finetuning on unoccluded images.

Abstract

Explore similar work

OASIC: Occlusion-Agnostic and Severity-Informed Classification

Robust Onion: Peeling Open Vocab Object Detectors Under Noise

Some Complexity Results for Robustness Verification for Binarized Neural Networks