Abstract:Knowledge-based Visual Question Answering (VQA) requires models to incorporate external knowledge to respond to questions about visual content. Previous methods mostly follow the "retrieve and generate" paradigm. Initially, they utilize a pre-trained retriever to fetch relevant knowledge documents, subsequently employing them to generate answers. While these methods have demonstrated commendable performance in the task, they possess limitations: (1) they employ an independent retriever to acquire knowledge solely based on the similarity between the query and knowledge embeddings, without assessing whether the knowledge document is truly conducive to helping answer the question; (2) they convert the image into text and then conduct retrieval and answering in natural language space, which may not ensure comprehensive acquisition of all image information. To address these limitations, we propose Boter, a novel framework designed to bootstrap knowledge selection and question answering by leveraging the robust multimodal perception capabilities of the Multimodal Large Language Model (MLLM). The framework consists of two modules: Selector and Answerer, where both are initialized by the MLLM and parameter-efficiently finetuned in a simple cycle: find key knowledge in the retrieved knowledge documents using the Selector, and then use them to finetune the Answerer to predict answers; obtain the pseudo-labels of key knowledge documents based on the predictions of the Answerer and weak supervision labels, and then finetune the Selector to select key knowledge; repeat. Our framework significantly enhances the performance of the baseline on the challenging open-domain Knowledge-based VQA benchmark, OK-VQA, achieving a state-of-the-art accuracy of 62.83%.
Abstract:Knowledge-based visual question answering (KB-VQA) is a challenging task, which requires the model to leverage external knowledge for comprehending and answering questions grounded in visual content. Recent studies retrieve the knowledge passages from external knowledge bases and then use them to answer questions. However, these retrieved knowledge passages often contain irrelevant or noisy information, which limits the performance of the model. To address the challenge, we propose two synergistic models: Knowledge Condensation model and Knowledge Reasoning model. We condense the retrieved knowledge passages from two perspectives. First, we leverage the multimodal perception and reasoning ability of the visual-language models to distill concise knowledge concepts from retrieved lengthy passages, ensuring relevance to both the visual content and the question. Second, we leverage the text comprehension ability of the large language models to summarize and condense the passages into the knowledge essence which helps answer the question. These two types of condensed knowledge are then seamlessly integrated into our Knowledge Reasoning model, which judiciously navigates through the amalgamated information to arrive at the conclusive answer. Extensive experiments validate the superiority of the proposed method. Compared to previous methods, our method achieves state-of-the-art performance on knowledge-based VQA datasets (65.1% on OK-VQA and 60.1% on A-OKVQA) without resorting to the knowledge produced by GPT-3 (175B).
Abstract:Vision and text have been fully explored in contemporary video-text foundational models, while other modalities such as audio and subtitles in videos have not received sufficient attention. In this paper, we resort to establish connections between multi-modality video tracks, including Vision, Audio, and Subtitle, and Text by exploring an automatically generated large-scale omni-modality video caption dataset called VAST-27M. Specifically, we first collect 27 million open-domain video clips and separately train a vision and an audio captioner to generate vision and audio captions. Then, we employ an off-the-shelf Large Language Model (LLM) to integrate the generated captions, together with subtitles and instructional prompts into omni-modality captions. Based on the proposed VAST-27M dataset, we train an omni-modality video-text foundational model named VAST, which can perceive and process vision, audio, and subtitle modalities from video, and better support various tasks including vision-text, audio-text, and multi-modal video-text tasks (retrieval, captioning and QA). Extensive experiments have been conducted to demonstrate the effectiveness of our proposed VAST-27M corpus and VAST foundation model. VAST achieves 22 new state-of-the-art results on various cross-modality benchmarks. Code, model and dataset will be released at https://github.com/TXH-mercury/VAST.