End-to-end Concept Word Detection for Video Captioning, Retrieval, and Question Answering

TL;DR

This paper introduces an end-to-end trainable concept word detector for video-to-language tasks that improves performance across multiple benchmarks without external knowledge sources.

Contribution

It presents a novel concept word detector integrated with video-to-language models, enhancing semantic understanding without external data and demonstrating superior results.

Findings

Achieved top accuracy in fill-in-the-blank, multiple-choice, and movie retrieval tasks.

Developed a semantic attention mechanism to focus on detected concept words.

Performed comparably on movie description task.

Abstract

We propose a high-level concept word detector that can be integrated with any video-to-language models. It takes a video as input and generates a list of concept words as useful semantic priors for language generation models. The proposed word detector has two important properties. First, it does not require any external knowledge sources for training. Second, the proposed word detector is trainable in an end-to-end manner jointly with any video-to-language models. To maximize the values of detected words, we also develop a semantic attention mechanism that selectively focuses on the detected concept words and fuse them with the word encoding and decoding in the language model. In order to demonstrate that the proposed approach indeed improves the performance of multiple video-to-language tasks, we participate in four tasks of LSMDC 2016. Our approach achieves the best accuracies in three of them, including fill-in-the-blank, multiple-choice test, and movie retrieval. We also attain comparable performance for the other task, movie description.