Agglomerative Token Clustering
Abstract
We present Agglomerative Token Clustering (ATC), a novel token merging method that consistently outperforms previous token merging and pruning methods across image classification, image synthesis, and object detection & segmentation tasks. ATC merges clusters through bottom-up hierarchical clustering, without the introduction of extra learnable parameters. We find that ATC achieves state-of-the-art performance across all tasks, and can even perform on par with prior state-of-the-art when applied off-the-shelf, i.e. without fine-tuning. ATC is particularly effective when applied with low keep rates, where only a small fraction of tokens are kept and retaining task performance is especially difficult.
Cite
@article{arxiv.2409.11923,
title = {Agglomerative Token Clustering},
author = {Joakim Bruslund Haurum and Sergio Escalera and Graham W. Taylor and Thomas B. Moeslund},
journal= {arXiv preprint arXiv:2409.11923},
year = {2024}
}
Comments
ECCV 2024. Project webpage at https://vap.aau.dk/atc/