Dual Hierarchical Temporal Convolutional Network with QA-Aware Dynamic Normalization for Video Story Question Answering

Abstract

Video story question answering (video story QA) is a challenging problem, as it requires a joint understanding of diverse data sources (i.e., video, subtitle, question, and answer choices). Existing approaches for video story QA have several common defects: (1) single temporal scale; (2) static and rough multimodal interaction; and (3) insufficient (or shallow) exploitation of both question and answer choices. In this paper, we propose a novel framework named Dual Hierarchical Temporal Convolutional Network (DHTCN) to address the aforementioned defects together. The proposed DHTCN explores multiple temporal scales by building hierarchical temporal convolutional network. In each temporal convolutional layer, two key components, namely AttLSTM and QA-Aware Dynamic Normalization, are introduced to capture the temporal dependency and the multimodal interaction in a dynamic and fine-grained manner. To enable sufficient exploitation of both question and answer choices, we increase the depth of QA pairs with a stack of nonlinear layers, and exploit QA pairs in each layer of the network. Extensive experiments are conducted on two widely used datasets: TVQA and MovieQA, demonstrating the effectiveness of DHTCN. Our model obtains state-of-the-art results on the both datasets. CCS CONCEPTS • Computing methodologies → Natural language processing; Computer vision.

Temporal Pyramid Transformer with Multimodal Interaction for Video Question Answering

“…Following previous studies (Li et al 2019b;Huang et al 2020;Fan et al 2019;Le et al 2020;Park, Lee, and Sohn 2021), different decoding methods are used according to the types of question. Specifically, we approach an open-ended question as a multi-class classification task, where the answering decoder aims to predict the correct class from the answer space A.…”

“…Most existing methods (Lei et al 2018;Gao et al 2019;Liu et al 2020;Cai et al 2020;Jiang and Han 2020) use recurrent neural networks (RNNs) and their Figure 1: The interaction of the question and the visual content usually happens at multiple temporal scales, as illustrated by the connected pairs of different parts of the question and frames at different levels of the temporal pyramid.…”