Ask, Attend and Answer: Exploring Question-Guided Spatial Attention for Visual Question Answering

Abstract

We address the problem of Visual Question Answering (VQA), which requires joint image and language understanding to answer a question about a given photograph. Recent approaches have applied deep image captioning methods based on convolutional-recurrent networks to this problem, but have failed to model spatial inference. To remedy this, we propose a model we call the Spatial Memory Network and apply it to the VQA task. Memory networks are recurrent neural networks with an explicit attention mechanism that selects certain parts of the information stored in memory. Our Spatial Memory Network stores neuron activations from different spatial regions of the image in its memory, and uses the question to choose relevant regions for computing the answer, a process of which constitutes a single "hop" in the network. We propose a novel spatial attention architecture that aligns words with image patches in the first hop, and obtain improved results by adding a second attention hop which considers the whole question to choose visual evidence based on the results of the first hop. To better understand the inference process learned by the network, we design synthetic questions that specifically require spatial inference and visualize the attention weights. We evaluate our model on two published visual question answering datasets, DAQUAR [1] and VQA [2], and obtain improved results compared to a strong deep baseline model (iBOWIMG) which concatenates image and question features to predict the answer [3].

include test-standard result on VQA full release (V1.0) dataset

References in corpus (4)

• Caffe: Convolutional Architecture for Fast Feature Embedding
• Simple Baseline for Visual Question Answering
• Question Answering with Subgraph Embeddings
• Ask Me Anything: Free-form Visual Question Answering Based on Knowledge from External Sources

Cited by in corpus (41)

• Hierarchical Question-Image Co-Attention for Visual Question Answering
• Understanding Neural Networks through Representation Erasure
• Modulating early visual processing by language
• Simple Baseline for Visual Question Answering
• Transparency by Design: Closing the Gap Between Performance and Interpretability in Visual Reasoning
• Learn to Combine Modalities in Multimodal Deep Learning
• A Focused Dynamic Attention Model for Visual Question Answering
• Multi-modal Factorized Bilinear Pooling with Co-Attention Learning for Visual Question Answering
• Bottom-Up and Top-Down Attention for Image Captioning and Visual Question Answering
• HydraPlus-Net: Attentive Deep Features for Pedestrian Analysis
• Attention Correctness in Neural Image Captioning
• Visual Question Answering: A Survey of Methods and Datasets
• Incorporating External Knowledge to Answer Open-Domain Visual Questions with Dynamic Memory Networks
• Implicit Distortion and Fertility Models for Attention-based Encoder-Decoder NMT Model
• Creativity: Generating Diverse Questions using Variational Autoencoders
• Segmentation from Natural Language Expressions
• Neural Attention Models for Sequence Classification: Analysis and Application to Key Term Extraction and Dialogue Act Detection
• Motion-Appearance Co-Memory Networks for Video Question Answering
• An Analysis of Visual Question Answering Algorithms
• Multi-step Joint-Modality Attention Network for Scene-Aware Dialogue System
• Utilizing Large Scale Vision and Text Datasets for Image Segmentation from Referring Expressions
• Asking the Difficult Questions: Goal-Oriented Visual Question Generation via Intermediate Rewards
• DualNet: Domain-Invariant Network for Visual Question Answering
• Learning Visual Knowledge Memory Networks for Visual Question Answering
• Factor Graph Attention
• Tell-and-Answer: Towards Explainable Visual Question Answering using Attributes and Captions
• Aspect-augmented Adversarial Networks for Domain Adaptation
• An Empirical Evaluation of Visual Question Answering for Novel Objects
• A Simple Loss Function for Improving the Convergence and Accuracy of Visual Question Answering Models
• MarioQA: Answering Questions by Watching Gameplay Videos
• Attend in groups: a weakly-supervised deep learning framework for learning from web data
• Learning to Disambiguate by Asking Discriminative Questions
• Query-by-example Spoken Term Detection using Attention-based Multi-hop Networks
• Mean Box Pooling: A Rich Image Representation and Output Embedding for the Visual Madlibs Task
• Spatial Memory for Context Reasoning in Object Detection
• Sparsely Grouped Multi-task Generative Adversarial Networks for Facial Attribute Manipulation
• Locally Smoothed Neural Networks
• Learning Models for Actions and Person-Object Interactions with Transfer to Question Answering
• Neural Rejuvenation: Improving Deep Network Training by Enhancing Computational Resource Utilization
• Deep Multi-Modal Sets
• Traversing the Continuous Spectrum of Image Retrieval with Deep Dynamic Models