Large Multimodal Models (LMMs) have made significant strides in visual question-answering for single images. Recent advancements like long-context LMMs have allowed them to ingest larger, or even multiple, images. However, the ability to process a large number of visual tokens does not guarantee effective retrieval and reasoning for multi-image question answering (MIQA), especially in real-world applications like photo album searches or satellite imagery analysis. In this work, we first assess the limitations of current benchmarks for long-context LMMs. We address these limitations by introducing a new vision-centric, long-context benchmark, "Visual Haystacks (VHs)". We comprehensively evaluate both open-source and proprietary models on VHs, and demonstrate that these models struggle when reasoning across potentially unrelated images, perform poorly on cross-image reasoning, as well as exhibit biases based on the placement of key information within the context window. Towards a solution, we introduce MIRAGE (Multi-Image Retrieval Augmented Generation), an open-source, lightweight visual-RAG framework that processes up to 10k images on a single 40G A100 GPU -- far surpassing the 1k-image limit of contemporary models. MIRAGE demonstrates up to 13% performance improvement over existing open-source LMMs on VHs, sets a new state-of-the-art on the RetVQA multi-image QA benchmark, and achieves competitive performance on single-image QA with state-of-the-art LMMs.
Videos, with their unique temporal dimension, demand precise grounded understanding, where answers are directly linked to visual, interpretable evidence. Despite significant breakthroughs in reasoning capabilities within Large Language Models, multi-modal reasoning - especially for videos - remains unexplored. In this work, we introduce VideoMind, a novel video-language agent designed for temporal-grounded video understanding. VideoMind incorporates two key innovations: (i) We identify essential capabilities for video temporal reasoning and develop a role-based agentic workflow, including a planner for coordinating different roles, a grounder for temporal localization, a verifier to assess temporal interval accuracy, and an answerer for question-answering. (ii) To efficiently integrate these diverse roles, we propose a novel Chain-of-LoRA strategy, enabling seamless role-switching via lightweight LoRA adaptors while avoiding the overhead of multiple models, thus balancing efficiency and flexibility. Extensive experiments on 14 public benchmarks, including 3 on grounded video question-answering (Grounded VideoQA), 6 on video temporal grounding (VTG), and 5 on general video question-answering (VideoQA), verify that our agent achieves state-of-the-art performance on diverse video understanding tasks, underscoring its effectiveness in advancing video agent and long-form temporal reasoning.
Natural language is ambiguous. Resolving ambiguous questions is key to successfully answering them. Focusing on questions about images, we create a dataset of ambiguous examples. We annotate these, grouping answers by the underlying question they address and rephrasing the question for each group to reduce ambiguity. Our analysis reveals a linguistically-aligned ontology of reasons for ambiguity in visual questions. We then develop an English question-generation model which we demonstrate via automatic and human evaluation produces less ambiguous questions. We further show that the question generation objective we use allows the model to integrate answer group information without any direct supervision.
We propose Encyclopedic-VQA, a large scale visual question answering (VQA) dataset featuring visual questions about detailed properties of fine-grained categories and instances. It contains 221k unique question+answer pairs each matched with (up to) 5 images, resulting in a total of 1M VQA samples. Moreover, our dataset comes with a controlled knowledge base derived from Wikipedia, marking the evidence to support each answer. Empirically, we show that our dataset poses a hard challenge for large vision+language models as they perform poorly on our dataset: PaLI [14] is state-of-the-art on OK-VQA [37], yet it only achieves 13.0% accuracy on our dataset. Moreover, we experimentally show that progress on answering our encyclopedic questions can be achieved by augmenting large models with a mechanism that retrieves relevant information from the knowledge base. An oracle experiment with perfect retrieval achieves 87.0% accuracy on the single-hop portion of our dataset, and an automatic retrieval-augmented prototype yields 48.8%. We believe that our dataset enables future research on retrieval-augmented vision+language models. It is available at https://github.com/google-research/google-research/tree/master/encyclopedic_vqa .
Recently a number of studies demonstrated impressive performance on diverse vision-language multi-modal tasks such as image captioning and visual question answering by extending the BERT architecture with multi-modal pre-training objectives. In this work we explore a broad set of multi-modal representation learning tasks in the medical domain, specifically using radiology images and the unstructured report. We propose Medical Vision Language Learner (MedViLL), which adopts a BERT-based architecture combined with a novel multi-modal attention masking scheme to maximize generalization performance for both vision-language understanding tasks (diagnosis classification, medical image-report retrieval, medical visual question answering) and vision-language generation task (radiology report generation). By statistically and rigorously evaluating the proposed model on four downstream tasks with three radiographic image-report datasets (MIMIC-CXR, Open-I, and VQA-RAD), we empirically demonstrate the superior downstream task performance of MedViLL against various baselines, including task-specific architectures. The source code is publicly available at: https://github.com/SuperSupermoon/MedViLL
Recent advancements in vision-language systems have improved the accuracy of Radiological Visual Question Answering (VQA) Models. However, some challenges remain across each stage of model development: limited expert-labeled images hinders data procurement at scale; the intricate and nuanced patterns of radiological images make modeling inherently difficult; and the lack of evaluation evaluation efforts makes it difficult to identify cases where the model might be ill-conditioned. In this study, we fine-tune a lightweight 3B parameter vision-language model for Radiological VQA, demonstrating that small models, when appropriately tuned with curated data, can achieve robust performance across both open- and closed-ended questions. We propose a cost-effective training pipeline from synthetic question-answer pair generation to multi-stage fine-tuning on specialised radiological domain-targeted datasets (e.g., ROCO v2.0, MedPix v2.0). Our results show that despite operating at a fraction of the scale of state-of-the-art models such as LLaVA-Med, our model achieves promising performance given its small parameter size and the limited scale of training data. We introduce a lightweight saliency-based diagnostic tool that enables domain experts to inspect VQA model performance and identify ill-conditioned failure modes through saliency analysis.
Visual Question-Answering (VQA) has become key to user experience, particularly after improved generalization capabilities of Vision-Language Models (VLMs). But evaluating VLMs for an application requirement using a standardized framework in practical settings is still challenging. This paper aims to solve that using an end-to-end framework. We present VQA360 - a novel dataset derived from established VQA benchmarks, annotated with task types, application domains, and knowledge types, for a comprehensive evaluation. We also introduce GoEval, a multimodal evaluation metric developed using GPT-4o, achieving a correlation factor of 56.71% with human judgments. Our experiments with state-of-the-art VLMs reveal that no single model excels universally, thus, making a right choice a key design decision. Proprietary models such as Gemini-1.5-Pro and GPT-4o-mini generally outperform others, but open-source models like InternVL-2-8B and CogVLM-2-Llama-3-19B also demonstrate competitive strengths, while providing additional advantages. Our framework can also be extended to other tasks.
다중 모달 대규모 언어 모델(Multimodal Large Language Models, MLLMs)의 공간적 사고 능력을 평가하기 위해 새로운 비디오 기반 시각-공간 지능 벤치마크(VSI-Bench)가 개발되었다. 이 벤치마크는 5,000개 이상의 질문-답변 쌍으로 구성되어 있으며, MLLMs가 인간 수준에는 미치지 못하지만 경쟁력 있는 시각-공간 지능을 보유하고 있음을 보여준다. 연구진은 모델이 공간을 어떻게 사고하는지를 언어적 및 시각적으로 탐구하였고, 공간 추론 능력이 MLLMs의 성능 향상의 주요 병목 현상으로 작용하고 있음을 발견하였다.
VSI-Bench의 구성은 데이터셋 수집 및 통합, 메타 정보 통합, 질문-답변 쌍 생성, 그리고 인간 검토 품질 관리의 단계로 이루어져 있다. 데이터셋은 ARKitScenes와 ScanNet에서 수집된 비디오를 기반으로 하며, 각 장면에 대한 표준화된 메타 정보가 포함되어 있다. 질문-답변 쌍은 다양한 작업 유형에 따라 생성되며, 품질 관리를 위해 인간 평가자가 참여한다.
MLLM의 사고 분석 과정에서는 자기 설명(self-explanations)과 오류 분석이 포함된다. 모델의 예측을 뒷받침하는 추론 체인을 명시적으로 추출하여 오류를 분석하며, 이를 통해 시각적 인식 오류, 언어적 지능 오류, 관계적 추론 오류 등의 다양한 오류 유형을 분류하였다. 또한, 인지 지도(cognitive maps)를 생성하여 MLLM의 공간 인식을 평가하는 방법이 제시되었다. 이 과정에서는 각 카테고리의 중심 위치를 예측하도록 모델을 프롬프트하고, 예측된 거리와 실제 거리 간의 차이를 정량적으로 평가한다.
이 연구는 MLLMs의 공간적 사고 능력을 심층적으로 분석하고, 기존의 언어적 추론 기법이 성능 향상에 실패하는 반면 인지 지도를 명시적으로 생성하는 것이 MLLMs의 공간 거리 능력을 향상시킨다는 중요한 발견을 제시한다. 이러한 결과는 MLLMs의 시각-공간 지능의 한계를 이해하는 데 기여하며, 향후 연구 방향에 대한 통찰을 제공한다.
Large language models such as GPT-3 have demonstrated an impressive capability to adapt to new tasks without requiring task-specific training data. This capability has been particularly effective in settings such as narrative question answering, where the diversity of tasks is immense, but the available supervision data is small. In this work, we investigate if such language models can extend their zero-shot reasoning abilities to long multimodal narratives in multimedia content such as drama, movies, and animation, where the story plays an essential role. We propose Long Story Short, a framework for narrative video QA that first summarizes the narrative of the video to a short plot and then searches parts of the video relevant to the question. We also propose to enhance visual matching with CLIPCheck. Our model outperforms state-of-the-art supervised models by a large margin, highlighting the potential of zero-shot QA for long videos.
Multimodal LLMs (MLLMs) are the natural extension of large language models to handle multimodal inputs, combining text and image data. They have recently garnered attention due to their capability to address complex tasks involving both modalities. However, their effectiveness is limited to the knowledge acquired during training, which restricts their practical utility. In this work, we introduce a novel method to enhance the adaptability of MLLMs by integrating external knowledge sources. Our proposed model, Reflective LLaVA (ReflectiVA), utilizes reflective tokens to dynamically determine the need for external knowledge and predict the relevance of information retrieved from an external database. Tokens are trained following a two-stage two-model training recipe. This ultimately enables the MLLM to manage external knowledge while preserving fluency and performance on tasks where external knowledge is not needed. Through our experiments, we demonstrate the efficacy of ReflectiVA for knowledge-based visual question answering, highlighting its superior performance compared to existing methods. Source code and trained models are publicly available at https://aimagelab.github.io/ReflectiVA.
In this work, we propose an efficient Video-Language Alignment (ViLA) network. Our ViLA model addresses both efficient frame sampling and effective cross-modal alignment in a unified way. In our ViLA network, we design a new learnable text-guided Frame-Prompter together with a new cross-modal distillation (QFormer-Distiller) module. Pre-trained large image-language models have shown promising results on problems such as visual question answering (VQA). However, how to efficiently and effectively sample video frames when adapting pre-trained large image-language model to video-language alignment is still the major challenge. Compared with prior work, our ViLA model demonstrates the capability of selecting key frames with critical contents, thus improving the video-language alignment accuracy while reducing the inference latency +3.3% on NExT-QA Temporal with 3.0X speed up). Overall, our ViLA network outperforms the state-of-the-art methods on the video question-answering benchmarks: +4.6% on STAR Interaction, +2.2% on STAR average with 3.0X speed up, ours 2-frames out-perform SeViLA 4-frames on the VLEP dataset with 4.2X speed-up. The code will be available at https://github.com/xijun-cs/ViLA.
Multiple works have emerged to push the boundaries on multi-modal large language models (MLLMs) towards pixel-level understanding. Such approaches have shown strong performance on benchmarks for referring expression segmentation and grounded conversation generation. The current trend in pixel-level MLLMs is to train with pixel-level grounding supervision on large-scale labelled data. However, we show that such MLLMs when evaluated on recent challenging vision centric benchmarks, exhibit a weak ability in visual question answering. Surprisingly, some of these methods even downgrade the grounding ability of MLLMs that were never trained with such supervision. In this work, we propose two novel challenging benchmarks and show that MLLMs without pixel-level grounding supervision can outperform the state of the art in such tasks when evaluating both the pixel-level grounding and visual question answering. We propose simple baselines to extract the grounding information that can be plugged into any MLLM, which we call as PixFoundation. More importantly, we study the research question of ``When does grounding emerge in MLLMs that are not trained with pixel-level grounding supervision?'' We show that grounding can coincide with object parts or location/appearance information. Code repository is at https://github.com/MSiam/PixFoundation/.
Recent methods for visual question answering rely on large-scale annotated datasets. Manual annotation of questions and answers for videos, however, is tedious, expensive and prevents scalability. In this work, we propose to avoid manual annotation and generate a large-scale training dataset for video question answering making use of automatic cross-modal supervision. We leverage a question generation transformer trained on text data and use it to generate question-answer pairs from transcribed video narrations. Given narrated videos, we then automatically generate the HowToVQA69M dataset with 69M video-question-answer triplets. To handle the open vocabulary of diverse answers in this dataset, we propose a training procedure based on a contrastive loss between a video-question multi-modal transformer and an answer transformer. We introduce the zero-shot VideoQA task and show excellent results, in particular for rare answers. Furthermore, we demonstrate our method to significantly outperform the state of the art on MSRVTT-QA, MSVD-QA, ActivityNet-QA and How2QA. Finally, for a detailed evaluation we introduce iVQA, a new VideoQA dataset with reduced language biases and high-quality redundant manual annotations. Our code, datasets and trained models are available at https://antoyang.github.io/just-ask.html.
Knowledge-based visual question answering (VQA) involves questions that require world knowledge beyond the image to yield the correct answer. Large language models (LMs) like GPT-3 are particularly helpful for this task because of their strong knowledge retrieval and reasoning capabilities. To enable LM to understand images, prior work uses a captioning model to convert images into text. However, when summarizing an image in a single caption sentence, which visual entities to describe are often underspecified. Generic image captions often miss visual details essential for the LM to answer visual questions correctly. To address this challenge, we propose PromptCap (Prompt-guided image Captioning), a captioning model designed to serve as a better connector between images and black-box LMs. Different from generic captions, PromptCap takes a natural-language prompt to control the visual entities to describe in the generated caption. The prompt contains a question that the caption should aid in answering. To avoid extra annotation, PromptCap is trained by examples synthesized with GPT-3 and existing datasets. We demonstrate PromptCap's effectiveness on an existing pipeline in which GPT-3 is prompted with image captions to carry out VQA. PromptCap outperforms generic captions by a large margin and achieves state-of-the-art accuracy on knowledge-based VQA tasks (60.4% on OK-VQA and 59.6% on A-OKVQA). Zero-shot results on WebQA show that PromptCap generalizes well to unseen domains.
Recent studies on machine reading comprehension have focused on text-level understanding but have not yet reached the level of human understanding of the visual layout and content of real-world documents. In this study, we introduce a new visual machine reading comprehension dataset, named VisualMRC, wherein given a question and a document image, a machine reads and comprehends texts in the image to answer the question in natural language. Compared with existing visual question answering (VQA) datasets that contain texts in images, VisualMRC focuses more on developing natural language understanding and generation abilities. It contains 30,000+ pairs of a question and an abstractive answer for 10,000+ document images sourced from multiple domains of webpages. We also introduce a new model that extends existing sequence-to-sequence models, pre-trained with large-scale text corpora, to take into account the visual layout and content of documents. Experiments with VisualMRC show that this model outperformed the base sequence-to-sequence models and a state-of-the-art VQA model. However, its performance is still below that of humans on most automatic evaluation metrics. The dataset will facilitate research aimed at connecting vision and language understanding.
Recent advancements in deep learning have led to the development of powerful language models (LMs) that excel in various tasks. Despite these achievements, there is still room for improvement, particularly in enhancing reasoning abilities and incorporating multimodal data. This report investigates the potential impact of combining Chain-of-Thought (CoT) reasoning and Visual Question Answering (VQA) techniques to improve LM's accuracy in solving multiple-choice questions. By employing TextVQA and ScienceQA datasets, we assessed the effectiveness of three text embedding methods and three visual embedding approaches. Our experiments aimed to fill the gap in current research by investigating the combined impact of CoT and VQA, contributing to the understanding of how these techniques can improve the reasoning capabilities of state-of-the-art models like GPT-4. Results from our experiments demonstrated the potential of these approaches in enhancing LM's reasoning and question-answering capabilities, providing insights for further research and development in the field, and paving the way for more accurate and reliable AI systems that can handle complex reasoning tasks across multiple modalities.
비디오 질문 응답(Video Question Answering, VQA) 시스템은 비디오 콘텐츠의 시각적, 시간적, 언어적 정보를 통합하여 질문에 대한 정확한 답변을 생성하는 복잡한 작업입니다. 기존의 VQA 접근법은 주로 프레임 수준의 캡션을 단일 모델에 입력하는 방식으로, 이는 시간적 동적 및 상호작용적 맥락을 충분히 반영하지 못하는 한계를 지니고 있습니다. 이러한 문제를 해결하기 위해 제안된 VideoMultiAgents 프레임워크는 비전, 장면 그래프 분석, 텍스트 처리를 위한 전문화된 에이전트를 통합하여 각 모달리티의 독립적인 추론을 통해 비디오 이해를 강화합니다.
VideoMultiAgents는 세 가지 전문 에이전트(텍스트 분석, 비디오 분석, 그래프 분석)로 구성되어 있으며, 각 에이전트는 독립적으로 작동하여 비디오의 다양한 측면을 분석하고 부분적인 답변을 생성합니다. 이 과정에서 조직자 에이전트는 여러 에이전트의 출력을 통합하여 일관된 최종 답변을 도출합니다. 또한, 질문 기반 캡션 생성 기법을 통해 주어진 질문과 밀접하게 관련된 객체, 행동 및 시간적 전환을 강조하는 캡션을 생성함으로써 답변의 정확성을 더욱 높입니다.
실험 결과, VideoMultiAgents는 Intent-QA, EgoSchema, NExT-QA 데이터셋에서 각각 79.0%, 75.4%, 79.6%의 정확도를 기록하며 기존의 최첨단 성능을 초과하는 성과를 달성했습니다. 이러한 성과는 다중 에이전트 아키텍처의 효과성과 질문 기반 캡션 생성의 중요성을 입증하며, VQA 분야에서의 새로운 가능성을 제시합니다. VideoMultiAgents는 비디오-언어 이해의 발전을 위한 유망한 방향을 제시하며, 향후 다양한 작업과 모달리티에 대한 확장 가능성을 가지고 있습니다.
Multimodal Large Language Models (MLLMs) have recently shown remarkable perceptual capability in answering visual questions, however, little is known about the limits of their perception. In particular, while prior works have provided anecdotal evidence of MLLMs' sensitivity to object size, this phenomenon and its underlying causes have not been explored comprehensively. In this work, we quantitatively study the perception of small visual objects in several state-of-the-art MLLMs and reveal a pervasive limitation in answering questions about small objects in images. Next, we identify four independent factors that can contribute to this limitation -- object quality, size, distractors, and location -- and conduct controlled intervention studies to measure the effect of each factor on MLLMs' perception. In particular, we find that lower object quality and smaller object size can both independently reduce MLLMs' ability to answer visual questions. More surprisingly, we find that the location of the object in the image and the presence of visual distractors can also significantly reduce MLLMs' question answering accuracy. Our study provides a better understanding of the perceptual limitation of MLLMs and contributes new evaluation protocols for analyzing the perception of future MLLMs. To facilitate further investigations, we release our code and data.
Video question answering (VideoQA) is challenging as it requires modeling capacity to distill dynamic visual artifacts and distant relations and to associate them with linguistic concepts. We introduce a general-purpose reusable neural unit called Conditional Relation Network (CRN) that serves as a building block to construct more sophisticated structures for representation and reasoning over video. CRN takes as input an array of tensorial objects and a conditioning feature, and computes an array of encoded output objects. Model building becomes a simple exercise of replication, rearrangement and stacking of these reusable units for diverse modalities and contextual information. This design thus supports high-order relational and multi-step reasoning. The resulting architecture for VideoQA is a CRN hierarchy whose branches represent sub-videos or clips, all sharing the same question as the contextual condition. Our evaluations on well-known datasets achieved new SoTA results, demonstrating the impact of building a general-purpose reasoning unit on complex domains such as VideoQA.
Encouraging progress has been made towards Visual Question Answering (VQA) in recent years, but it is still challenging to enable VQA models to adaptively generalize to out-of-distribution (OOD) samples. Intuitively, recompositions of existing visual concepts (\ie, attributes and objects) can generate unseen compositions in the training set, which will promote VQA models to generalize to OOD samples. In this paper, we formulate OOD generalization in VQA as a compositional generalization problem and propose a graph generative modeling-based training scheme (X-GGM) to implicitly model the problem. X-GGM leverages graph generative modeling to iteratively generate a relation matrix and node representations for the predefined graph that utilizes attribute-object pairs as nodes. Furthermore, to alleviate the unstable training issue in graph generative modeling, we propose a gradient distribution consistency loss to constrain the data distribution with adversarial perturbations and the generated distribution. The baseline VQA model (LXMERT) trained with the X-GGM scheme achieves state-of-the-art OOD performance on two standard VQA OOD benchmarks, \ie, VQA-CP v2 and GQA-OOD. Extensive ablation studies demonstrate the effectiveness of X-GGM components. Code is available at \url{https://github.com/jingjing12110/x-ggm}.