AI Paper Atlas

Research map / Architectures and efficiency / CNN design

FOVI: A biologically-inspired foveated interface for deep vision models

Nicholas Blauch, George A. Alvarez, Talia Konkle

ICML 2026 · Architectures and efficiency › CNN design · approach: Dense matching and estimation CNNs · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use foveated kNN-convolution patch embedding (a=2.79) instead of uniform downsampling to 64x64DINOv3 ViT-S+ @ 64, 1 fixation · ImageNet-1K valtop-1 accuracy: 0.693 -> 0.700Improved
use foveated sampling instead of log-polar samplingDINOv3 ViT-S+ @ 64, 1 fixation · ImageNet-1K valtop-1 accuracy: 0.643 -> 0.700Improved
increase foveation from weak (a=60.94) to a=2.79 (ablation)DINOv3 ViT-S+ @ 64, 1 fixation · ImageNet-1K valtop-1 accuracy: 0.693 -> 0.700Improved
increase foveation from weak (a=60.94) to a=2.79 (ablation)DINOv3 ViT-S+ @ 64, 3 fixations · ImageNet-1K valtop-1 accuracy: 0.725 -> 0.735Improved
add two more fixations (1 -> 3)FOVI-ViT-H+ @ 64 · ImageNet-1K valtop-1 accuracy: 0.844 -> 0.853Improved
add two more fixations (1 -> 3)FOVI-ViT-S+ @ 64 · ImageNet-1K valtop-1 accuracy: 0.700 -> 0.735Improved
replace uniform @ 224 with FOVI @ 64 (1/16 pixels)DINOv3 ViT-H+ · ImageNet-1K valGFLOPS per image: 172.39 -> 58.43Improved
replace uniform @ 224 with FOVI @ 64 (1/16 pixels)DINOv3 ViT-H+ · ImageNet-1K valtop-1 accuracy: 0.871 -> 0.844Hurt
raise sensor resolution from 64 to 224 pixels per side with the same 64 patches (ablation)FOVI-ViT-S+ (1 fixation) · ImageNet-1K valtop-1 accuracy: 0.700 -> 0.737Improved
LoRA on patch embedding and first half of network instead of frozen adaptation (ablation)DINOv3 on foveated input · ImageNet-100accuracy: outperforms frozen adaptation by ~30%Improved
LoRA instead of full fine-tuning (ablation)DINOv3 on foveated input · ImageNet-100accuracy: outperforms full fine-tuning by ~10%Improved
adapt first half of the network instead of only late layers (ablation)DINOv3 on foveated input (LoRA) · ImageNet-100accuracy: outperforms late-only adaptation by ~15%Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy