Learning Patch-to-Cluster Attention in Vision Transformer

Abstract

The Vision Transformer (ViT) model is built on the assumption of treating image patches as "visual tokens" and learning patch-to-patch attention. The patch embedding based tokenizer is a workaround in practice and has a semantic gap with respect to its counterpart, the textual tokenizer. The patch-to-patch attention suffers from the quadratic complexity issue, and also makes it non-trivial to explain learned ViT models. To address these issues in ViT models, this paper proposes to learn patch-to-cluster attention (PaCa) based ViT models. Queries in our PaCa-ViT are based on patches, while keys and values are based on clustering (with a predefined small number of clusters). The clusters are learned end-to-end, leading to better tokenizers and realizing joint clustering-for-attention and attention-for-clustering when deployed in ViT models. The quadratic complexity is relaxed to linear complexity. Also, directly visualizing the learned clusters can reveal how a trained ViT model learns to perform a task (e.g., object detection). In experiments, the proposed PaCa-ViT is tested on CIFAR-100 and ImageNet-1000 image classification, and MS-COCO object detection and instance segmentation. Compared with prior arts, it obtains better performance in classification and comparable performance in detection and segmentation. It is significantly more efficient in COCO due to the linear complexity. The learned clusters are also semantically meaningful and shed light on designing more discriminative yet interpretable ViT models.

Scale Enhancement Pyramid Network for Small Object Detection from UAV Images

“…The attention mechanism is recognized as a potential means to enhance deep CNNs since it allows the network to selectively focus on the most important regions of an image while ignoring the ones with irrelevant parts. Currently, attention mechanisms are prevalent in various tasks, such as machine translation [ 42 ], object detection [ 43 ], and semantic segmentation [ 44 ]. More recently, multiple attention mechanisms have provided benefits in visual studies to improve convolutional network expression ability.…”