AI Navigate

Bioinspired CNNs for border completion in occluded images

arXiv cs.CV / 3/12/2026

📰 NewsModels & Research
Read original →
共有:

Key Points

  • The authors design BorderNet, a CNN whose filters are inspired by border-completion mechanisms in the visual cortex to improve robustness to occlusions.
  • They evaluate BorderNet on occluded MNIST, Fashion-MNIST, and EMNIST using stripe and grid occlusions, reporting improved performance over baselines with results that vary by occlusion severity and dataset.
  • The study demonstrates how neuroscience-inspired filter design can enhance occluded-object recognition in CNNs, suggesting potential for more robust vision systems.
  • This is an arXiv preprint (v1) announced as a new contribution in the field.

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.

関連おすすめサービス

※当サイトはアフィリエイト広告を利用しています
🎧

Notta搭載AI議事録イヤホン ZENCHORD1

AI時代の仕事術。Notta搭載で会議の議事録を自動生成するスマートイヤホン。

🎙️

AI搭載ボイスレコーダー Plaud

世界100万人が愛用。AIで文字起こし・要約を自動化するボイスレコーダー。

🖼️

画像高画質化AIツール Aiarty Image Enhancer

AIで画像を高画質化。写真・イラストを簡単にアップスケール。