Enhancing CNNs Robustness to Occlusions with Bioinspired Filters for Border Completion

会议论文
发表日期：

2026年

查看全文

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

摘要

We exploit the mathematical modeling of the visual cortex mechanism for border completion to define custom filters for CNNs. We see a consistent improvement in performance, particularly in accuracy, when our modified LeNet 5 is tested with occluded MNIST images.

We exploit the mathematical modeling of the visual cortex mechanism&#xa0;for border completion to define custom filters for CNNs. We&#xa0;see a consistent improvement in performance, particularly in accuracy, when&#xa0;our modified LeNet 5 is tested with occluded MNIST images.

Enhancing CNNs Robustness to Occlusions with Bioinspired Filters for Border Completion

摘要

We exploit the mathematical modeling of the visual cortex mechanism for border completion to define custom filters for CNNs. We see a consistent improvement in performance, particularly in accuracy, when our modified LeNet 5 is tested with occluded MNIST images.