SCA-CNN: Spatial and Channel-Wise Attention in Convolutional Networks for Image Captioning

TLDR

Visual attention has been applied to tasks such as image captioning and question answering, but existing models focus mainly on spatial probabilities and may not fully exploit the spatial, channel‑wise, and multi‑layer nature of CNN features. This paper introduces SCA‑CNN, a convolutional neural network that integrates spatial and channel‑wise attentions. SCA‑CNN dynamically modulates the sentence‑generation context across multi‑layer feature maps, attending to both spatial locations and channel activations, and is evaluated on Flickr8K, Flickr30K, and MSCOCO. The results show that SCA‑CNN consistently outperforms state‑of‑the‑art visual attention‑based image captioning methods.

Abstract

Visual attention has been successfully applied in structural prediction tasks such as visual captioning and question answering. Existing visual attention models are generally spatial, i.e., the attention is modeled as spatial probabilities that re-weight the last conv-layer feature map of a CNN encoding an input image. However, we argue that such spatial attention does not necessarily conform to the attention mechanism - a dynamic feature extractor that combines contextual fixations over time, as CNN features are naturally spatial, channel-wise and multi-layer. In this paper, we introduce a novel convolutional neural network dubbed SCA-CNN that incorporates Spatial and Channel-wise Attentions in a CNN. In the task of image captioning, SCA-CNN dynamically modulates the sentence generation context in multi-layer feature maps, encoding where (i.e., attentive spatial locations at multiple layers) and what (i.e., attentive channels) the visual attention is. We evaluate the proposed SCA-CNN architecture on three benchmark image captioning datasets: Flickr8K, Flickr30K, and MSCOCO. It is consistently observed that SCA-CNN significantly outperforms state-of-the-art visual attention-based image captioning methods.