IC-035Removing the inductive bias of locality from Vision Transformers improves or matches performance on classification and regression tasks.

Duy Kien Nguyen, Mido Assran, Unnat Jain, Martin R. Oswald, Cees G. M. Snoek, Xinlei Chen

SourceAn Image is Worth More Than 16x16 Patches: Exploring Transformers on Individual Pixels

The paper compares a standard Vision Transformer (ViT), which uses non-degenerate patches (e.g., 2x2), to a locality-free variant (ViT/1) that treats each individual pixel as a token with learnable position embeddings. This locality-free model is evaluated on CIFAR-100, ImageNet, Oxford-102-Flowers, and NYU-v2 depth estimation. The key finding is that the pixel-based transformer, which has zero prior about the 2D grid structure, consistently outperforms its patch-based counterpart across all these tasks, challenging the conventional belief that locality is a fundamental inductive bias for vision.

Evidence
correlational
Key metric
CIFAR-100: ViT-t/2 (83.6% acc@1) vs. ViT-t/1 (85.1%); ViT-s/2 (83.7%) vs. ViT-s/1 (86.4%). ImageNet: ViT-b/2 (75.7%) vs. ViT-b/1 (76.1%). Oxford-102-Flower: ViT-s/2 (45.8%) vs. ViT-s/1 (46.3%). NYU-v2 RMSE: ViT-s/2 (0.80) vs. ViT-s/1 (0.72).
Caveat
The paper notes that the pixel-based transformer is computationally expensive due to the quadratic complexity of self-attention, making it 'more of an approach for investigation, and less for applications.'
Model
ViT
Datasets
CIFAR-100 [eval], ImageNet-1k / ImageNet / ImageNet-1k-val / ImageNet-Val [eval], Oxford-102-Flowers [eval], NYU-v2 / NYU-D [eval]
Related work
ViT [builds-on]
Extraction
automatic-extraction