ALADIN: All Layer Adaptive Instance Normalization for Fine-grained Style Similarity

TL;DR

ALADIN is a new architecture that improves fine-grained style similarity search in images by leveraging a large-scale dataset and weak supervision, achieving state-of-the-art accuracy in visual search tasks.

Contribution

It introduces ALADIN, a novel All Layer AdaIN architecture, and a large-scale dataset BAM-FG for weakly supervised learning of fine-grained artistic style similarity.

Findings

Sets new state-of-the-art accuracy in style-based visual search.

Introduces BAM-FG, a dataset of 2.62 million images with 310,000 style groupings.

Demonstrates effectiveness of weakly supervised learning for style similarity.

Abstract

We present ALADIN (All Layer AdaIN); a novel architecture for searching images based on the similarity of their artistic style. Representation learning is critical to visual search, where distance in the learned search embedding reflects image similarity. Learning an embedding that discriminates fine-grained variations in style is hard, due to the difficulty of defining and labelling style. ALADIN takes a weakly supervised approach to learning a representation for fine-grained style similarity of digital artworks, leveraging BAM-FG, a novel large-scale dataset of user generated content groupings gathered from the web. ALADIN sets a new state of the art accuracy for style-based visual search over both coarse labelled style data (BAM) and BAM-FG; a new 2.62 million image dataset of 310,000 fine-grained style groupings also contributed by this work.