Multimodal conversational generative AI has shown impressive capabilities in various vision and language understanding through learning massive text-image data. However, current conversational models still lack knowledge about visual insects since they are often trained on the general knowledge of vision-language data. Meanwhile, understanding insects is a fundamental problem in precision agriculture, helping to promote sustainable development in agriculture. Therefore, this paper proposes a novel multimodal conversational model, Insect-LLaVA, to promote visual understanding in insect-domain knowledge. In particular, we first introduce a new large-scale Multimodal Insect Dataset with Visual Insect Instruction Data that enables the capability of learning the multimodal foundation models. Our proposed dataset enables conversational models to comprehend the visual and semantic features of the insects. Second, we propose a new Insect-LLaVA model, a new general Large Language and Vision Assistant in Visual Insect Understanding. Then, to enhance the capability of learning insect features, we develop an Insect Foundation Model by introducing a new micro-feature self-supervised learning with a Patch-wise Relevant Attention mechanism to capture the subtle differences among insect images. We also present Description Consistency loss to improve micro-feature learning via text descriptions. The experimental results evaluated on our new Visual Insect Question Answering benchmarks illustrate the effective performance of our proposed approach in visual insect understanding and achieve State-of-the-Art performance on standard benchmarks of insect-related tasks.
Video causal reasoning aims to achieve a high-level understanding of videos from a causal perspective. However, it exhibits limitations in its scope, primarily executed in a question-answering paradigm and focusing on brief video segments containing isolated events and basic causal relations, lacking comprehensive and structured causality analysis for videos with multiple interconnected events. To fill this gap, we introduce a new task and dataset, Multi-Event Causal Discovery (MECD). It aims to uncover the causal relations between events distributed chronologically across long videos. Given visual segments and textual descriptions of events, MECD identifies the causal associations between these events to derive a comprehensive and structured event-level video causal graph explaining why and how the result event occurred. To address the challenges of MECD, we devise a novel framework inspired by the Granger Causality method, incorporating an efficient mask-based event prediction model to perform an Event Granger Test. It estimates causality by comparing the predicted result event when premise events are masked versus unmasked. Furthermore, we integrate causal inference techniques such as front-door adjustment and counterfactual inference to mitigate challenges in MECD like causality confounding and illusory causality. Additionally, context chain reasoning is introduced to conduct more robust and generalized reasoning. Experiments validate the effectiveness of our framework in reasoning complete causal relations, outperforming GPT-4o and VideoChat2 by 5.77% and 2.70%, respectively. Further experiments demonstrate that causal relation graphs can also contribute to downstream video understanding tasks such as video question answering and video event prediction.
When trained at sufficient scale, auto-regressive language models exhibit the notable ability to learn a new language task after being prompted with just a few examples. Here, we present a simple, yet effective, approach for transferring this few-shot learning ability to a multimodal setting (vision and language). Using aligned image and caption data, we train a vision encoder to represent each image as a sequence of continuous embeddings, such that a pre-trained, frozen language model prompted with this prefix generates the appropriate caption. The resulting system is a multimodal few-shot learner, with the surprising ability to learn a variety of new tasks when conditioned on examples, represented as a sequence of multiple interleaved image and text embeddings. We demonstrate that it can rapidly learn words for new objects and novel visual categories, do visual question-answering with only a handful of examples, and make use of outside knowledge, by measuring a single model on a variety of established and new benchmarks.
Vision Language Models (VLMs) are typically evaluated with Visual Question Answering (VQA) tasks which assess a model's understanding of scenes. Good VQA performance is taken as evidence that the model will perform well on a broader range of tasks that require both visual and language inputs. However, scene-aware VQA does not fully capture input biases or assess hallucinations caused by a misalignment between modalities. To address this, we propose a Cross-modal Alignment Similarity Test (CAST) to probe VLMs for self-consistency across modalities. This test involves asking the models to identify similarities between two scenes through text-only, image-only, or both and then assess the truthfulness of the similarities they generate. Since there is no ground-truth to compare against, this evaluation does not focus on objective accuracy but rather on whether VLMs are internally consistent in their outputs. We argue that while not all self-consistent models are capable or accurate, all capable VLMs must be self-consistent.
Multimodal Large Language Models (MLLMs) have significantly advanced visual tasks by integrating visual representations into large language models (LLMs). The textual modality, inherited from LLMs, equips MLLMs with abilities like instruction following and in-context learning. In contrast, the visual modality enhances performance in downstream tasks by leveraging rich semantic content, spatial information, and grounding capabilities. These intrinsic modalities work synergistically across various visual tasks. Our research initially reveals a persistent imbalance between these modalities, with text often dominating output generation during visual instruction tuning. This imbalance occurs when using both full fine-tuning and parameter-efficient fine-tuning (PEFT) methods. We then found that re-balancing these modalities can significantly reduce the number of trainable parameters required, inspiring a direction for further optimizing visual instruction tuning. We introduce Modality Linear Representation-Steering (MoReS) to achieve the goal. MoReS effectively re-balances the intrinsic modalities throughout the model, where the key idea is to steer visual representations through linear transformations in the visual subspace across each model layer. To validate our solution, we composed LLaVA Steering, a suite of models integrated with the proposed MoReS method. Evaluation results show that the composed LLaVA Steering models require, on average, 500 times fewer trainable parameters than LoRA needs while still achieving comparable performance across three visual benchmarks and eight visual question-answering tasks. Last, we present the LLaVA Steering Factory, an in-house developed platform that enables researchers to quickly customize various MLLMs with component-based architecture for seamlessly integrating state-of-the-art models, and evaluate their intrinsic modality imbalance.
Dramatic progress has been witnessed in basic vision tasks involving low-level perception, such as object recognition, detection, and tracking. Unfortunately, there is still an enormous performance gap between artificial vision systems and human intelligence in terms of higher-level vision problems, especially ones involving reasoning. Earlier attempts in equipping machines with high-level reasoning have hovered around Visual Question Answering (VQA), one typical task associating vision and language understanding. In this work, we propose a new dataset, built in the context of Raven's Progressive Matrices (RPM) and aimed at lifting machine intelligence by associating vision with structural, relational, and analogical reasoning in a hierarchical representation. Unlike previous works in measuring abstract reasoning using RPM, we establish a semantic link between vision and reasoning by providing structure representation. This addition enables a new type of abstract reasoning by jointly operating on the structure representation. Machine reasoning ability using modern computer vision is evaluated in this newly proposed dataset. Additionally, we also provide human performance as a reference. Finally, we show consistent improvement across all models by incorporating a simple neural module that combines visual understanding and structure reasoning.
Advancements in Computer-Aided Screening (CAS) systems are essential for improving the detection of security threats in X-ray baggage scans. However, current datasets are limited in representing real-world, sophisticated threats and concealment tactics, and existing approaches are constrained by a closed-set paradigm with predefined labels. To address these challenges, we introduce STCray, the first multimodal X-ray baggage security dataset, comprising 46,642 image-caption paired scans across 21 threat categories, generated using an X-ray scanner for airport security. STCray is meticulously developed with our specialized protocol that ensures domain-aware, coherent captions, that lead to the multi-modal instruction following data in X-ray baggage security. This allows us to train a domain-aware visual AI assistant named STING-BEE that supports a range of vision-language tasks, including scene comprehension, referring threat localization, visual grounding, and visual question answering (VQA), establishing novel baselines for multi-modal learning in X-ray baggage security. Further, STING-BEE shows state-of-the-art generalization in cross-domain settings. Code, data, and models are available at https://divs1159.github.io/STING-BEE/.
We propose Pixel-BERT to align image pixels with text by deep multi-modal transformers that jointly learn visual and language embedding in a unified end-to-end framework. We aim to build a more accurate and thorough connection between image pixels and language semantics directly from image and sentence pairs instead of using region-based image features as the most recent vision and language tasks. Our Pixel-BERT which aligns semantic connection in pixel and text level solves the limitation of task-specific visual representation for vision and language tasks. It also relieves the cost of bounding box annotations and overcomes the unbalance between semantic labels in visual task and language semantic. To provide a better representation for down-stream tasks, we pre-train a universal end-to-end model with image and sentence pairs from Visual Genome dataset and MS-COCO dataset. We propose to use a random pixel sampling mechanism to enhance the robustness of visual representation and to apply the Masked Language Model and Image-Text Matching as pre-training tasks. Extensive experiments on downstream tasks with our pre-trained model show that our approach makes the most state-of-the-arts in downstream tasks, including Visual Question Answering (VQA), image-text retrieval, Natural Language for Visual Reasoning for Real (NLVR). Particularly, we boost the performance of a single model in VQA task by 2.17 points compared with SOTA under fair comparison.
The problem of realistic VQA (RVQA), where a model has to reject unanswerable questions (UQs) and answer answerable ones (AQs), is studied. We first point out 2 drawbacks in current RVQA research, where (1) datasets contain too many unchallenging UQs and (2) a large number of annotated UQs are required for training. To resolve the first drawback, we propose a new testing dataset, RGQA, which combines AQs from an existing VQA dataset with around 29K human-annotated UQs. These UQs consist of both fine-grained and coarse-grained image-question pairs generated with 2 approaches: CLIP-based and Perturbation-based. To address the second drawback, we introduce an unsupervised training approach. This combines pseudo UQs obtained by randomly pairing images and questions, with an RoI Mixup procedure to generate more fine-grained pseudo UQs, and model ensembling to regularize model confidence. Experiments show that using pseudo UQs significantly outperforms RVQA baselines. RoI Mixup and model ensembling further increase the gain. Finally, human evaluation reveals a performance gap between humans and models, showing that more RVQA research is needed.
상세한 비디오 캡션은 객체, 행동, 시간에 따른 사건을 빠짐없이 충실하게 묘사해야 한다. 그러나 프레임을 고정된 방식으로 샘플링하면 짧은 움직임, 작은 객체, 기타 세밀한 근거를 놓칠 수 있다. 따라서 관련 순간을 능동적으로 찾아 국소 영역을 살펴보는 캡션 에이전트가 필요하다. 비디오 질의응답과 달리 일반적인 캡션 작성 지시에는 어떤 시각적 근거를 확보해야 하는지 명시되지 않아 이러한 행동을 학습시키기 어렵다. 학습 시에만 제공하는 지각 질문(PQ)으로 비디오의 능동적 관찰을 유도하는 지각 질문 온-정책 증류(PQ-OPD)를 제안한다. 지각 질문은 무엇을 살펴봐야 하는지 알려 주면서 해당 답은 공개하지 않는다. 교사 모델이 비디오를 더 폭넓게 탐색하고 관련 시각적 근거에 집중하도록 유도한다. 학생 모델이 생성한 궤적에서 온-정책 증류를 적용해 이 지침을 학생의 추론, 도구 호출, 캡션에 전달한다. 추론 시에는 지각 질문이 필요하지 않다. 상세 캡션 벤치마크 세 개와 캡션 기반 질의응답 벤치마크 두 개에서, 교사와 학생의 규모 및 입력 프레임 예산이 달라도 PQ-OPD를 적용한 에이전트가 추가 학습 없이 사용한 에이전트보다 일관되게 우수했다. 이 평가 절차에서 PQ-OPD는 DREAM-1K, CaReBench, ShortVidBench, MotionBench, VidCapBench-AE의 주요 평가지표에서 비교한 공개 소스 모델 가운데 가장 높은 성능을 보였다. 구성 요소 제거 실험과 시간축 포괄 범위 및 답변 누출 분석을 통해 학생이 학습 시의 지침을 내재화해 더 효과적인 비디오 탐색 정책을 갖추게 됨을 확인했다.
Multi-modal Large language models (MLLMs) show remarkable ability in video understanding. Nevertheless, understanding long videos remains challenging as the models can only process a finite number of frames in a single inference, potentially omitting crucial visual information. To address the challenge, we propose generating multiple predictions through visual context sampling, followed by a scoring mechanism to select the final prediction. Specifically, we devise a bin-wise sampling strategy that enables MLLMs to generate diverse answers based on various combinations of keyframes, thereby enriching the visual context. To determine the final prediction from the sampled answers, we employ a self-reward by linearly combining three scores: (1) a frequency score indicating the prevalence of each option, (2) a marginal confidence score reflecting the inter-intra sample certainty of MLLM predictions, and (3) a reasoning score for different question types, including clue-guided answering for global questions and temporal self-refocusing for local questions. The frequency score ensures robustness through majority correctness, the confidence-aligned score reflects prediction certainty, and the typed-reasoning score addresses cases with sparse key visual information using tailored strategies. Experiments show that this approach covers the correct answer for a high percentage of long video questions, on seven datasets show that our method improves the performance of three MLLMs.
Visually-conditioned language models (VLMs) have seen growing adoption in applications such as visual dialogue, scene understanding, and robotic task planning; adoption that has fueled a wealth of new models such as LLaVa, InstructBLIP, and PaLI-3. Despite the volume of new releases, key design decisions around image preprocessing, architecture, and optimization are under-explored, making it challenging to understand what factors account for model performance $-$ a challenge further complicated by the lack of objective, consistent evaluations. To address these gaps, we first compile a suite of standardized evaluations spanning visual question answering, object localization, and challenge sets that probe properties such as hallucination; evaluations that provide fine-grained insight VLM capabilities. Second, we rigorously investigate VLMs along key design axes, including pretrained visual representations and training from base vs. instruct-tuned language models, amongst others. We couple our analysis with three resource contributions: (1) a unified framework for evaluating VLMs, (2) optimized, flexible training code, and (3) checkpoints for all models, including a family of VLMs at the 7-13B scale that strictly outperform InstructBLIP and LLaVa v1.5, the state-of-the-art in open VLMs.
Medical Visual Question Answering (Medical-VQA) aims to to answer clinical questions regarding radiology images, assisting doctors with decision-making options. Nevertheless, current Medical-VQA models learn cross-modal representations through residing vision and texture encoders in dual separate spaces, which lead to indirect semantic alignment. In this paper, we propose UnICLAM, a Unified and Interpretable Medical-VQA model through Contrastive Representation Learning with Adversarial Masking. Specifically, to learn an aligned image-text representation, we first establish a unified dual-stream pre-training structure with the gradually soft-parameter sharing strategy. Technically, the proposed strategy learns a constraint for the vision and texture encoders to be close in a same space, which is gradually loosened as the higher number of layers. Moreover, for grasping the unified semantic representation, we extend the adversarial masking data augmentation to the contrastive representation learning of vision and text in a unified manner. Concretely, while the encoder training minimizes the distance between original and masking samples, the adversarial masking module keeps adversarial learning to conversely maximize the distance. Furthermore, we also intuitively take a further exploration to the unified adversarial masking augmentation model, which improves the potential ante-hoc interpretability with remarkable performance and efficiency. Experimental results on VQA-RAD and SLAKE public benchmarks demonstrate that UnICLAM outperforms existing 11 state-of-the-art Medical-VQA models. More importantly, we make an additional discussion about the performance of UnICLAM in diagnosing heart failure, verifying that UnICLAM exhibits superior few-shot adaption performance in practical disease diagnosis.
Multi-modal large language models (MLLMs) have shown incredible capabilities in a variety of 2D vision and language tasks. We extend MLLMs' perceptual capabilities to ground and reason about images in 3-dimensional space. To that end, we first develop a large-scale pre-training dataset for 2D and 3D called LV3D by combining multiple existing 2D and 3D recognition datasets under a common task formulation: as multi-turn question-answering. Next, we introduce a new MLLM named Cube-LLM and pre-train it on LV3D. We show that pure data scaling makes a strong 3D perception capability without 3D specific architectural design or training objective. Cube-LLM exhibits intriguing properties similar to LLMs: (1) Cube-LLM can apply chain-of-thought prompting to improve 3D understanding from 2D context information. (2) Cube-LLM can follow complex and diverse instructions and adapt to versatile input and output formats. (3) Cube-LLM can be visually prompted such as 2D box or a set of candidate 3D boxes from specialists. Our experiments on outdoor benchmarks demonstrate that Cube-LLM significantly outperforms existing baselines by 21.3 points of AP-BEV on the Talk2Car dataset for 3D grounded reasoning and 17.7 points on the DriveLM dataset for complex reasoning about driving scenarios, respectively. Cube-LLM also shows competitive results in general MLLM benchmarks such as refCOCO for 2D grounding with (87.0) average score, as well as visual question answering benchmarks such as VQAv2, GQA, SQA, POPE, etc. for complex reasoning. Our project is available at https://janghyuncho.github.io/Cube-LLM.
We introduce MASSES, a simple evaluation metric for the task of Visual Question Answering (VQA). In its standard form, the VQA task is operationalized as follows: Given an image and an open-ended question in natural language, systems are required to provide a suitable answer. Currently, model performance is evaluated by means of a somehow simplistic metric: If the predicted answer is chosen by at least 3 human annotators out of 10, then it is 100% correct. Though intuitively valuable, this metric has some important limitations. First, it ignores whether the predicted answer is the one selected by the Majority (MA) of annotators. Second, it does not account for the quantitative Subjectivity (S) of the answers in the sample (and dataset). Third, information about the Semantic Similarity (SES) of the responses is completely neglected. Based on such limitations, we propose a multi-component metric that accounts for all these issues. We show that our metric is effective in providing a more fine-grained evaluation both on the quantitative and qualitative level.
FlameVQA는 FLAME 3의 UAV RGB와 방사형 열화상 TIFF를 결합한 다지선다형 VQA 벤치마크로, 검출·위치 추정·범위 추정·교차 모달 추론·비행 계획 등 6가지 산불 운용 역량을 이미지당 34문항으로 평가한다. MLLM 베이스라인에서는 교차 모달 단서가 분명할 때 성능이 높았으나, 짙은 연기 속 존재 검출과 커버리지 추정에서 실패가 뚜렷해 현행 MLLM은 재난·산불 모니터링을 위한 도메인 특화 적응이 필요함을 보여준다.
Recent progress in Large Vision-Language Models (LVLMs) has enabled promising applications in medical tasks, such as report generation and visual question answering. However, existing benchmarks focus mainly on the final diagnostic answer, offering limited insight into whether models engage in clinically meaningful reasoning. To address this, we present CheXStruct and CXReasonBench, a structured pipeline and benchmark built on the publicly available MIMIC-CXR-JPG dataset. CheXStruct automatically derives a sequence of intermediate reasoning steps directly from chest X-rays, such as segmenting anatomical regions, deriving anatomical landmarks and diagnostic measurements, computing diagnostic indices, and applying clinical thresholds. CXReasonBench leverages this pipeline to evaluate whether models can perform clinically valid reasoning steps and to what extent they can learn from structured guidance, enabling fine-grained and transparent assessment of diagnostic reasoning. The benchmark comprises 18,988 QA pairs across 12 diagnostic tasks and 1,200 cases, each paired with up to 4 visual inputs, and supports multi-path, multi-stage evaluation including visual grounding via anatomical region selection and diagnostic measurements. Even the strongest of 10 evaluated LVLMs struggle with structured reasoning and generalization, often failing to link abstract knowledge with anatomically grounded visual interpretation. The code is available at https://github.com/ttumyche/CXReasonBench
Recent advances in Vision-Language Models (VLMs) and the scarcity of high-quality multi-modal alignment data have inspired numerous researches on synthetic VLM data generation. The conventional norm in VLM data construction uses a mixture of specialists in caption and OCR, or stronger VLM APIs and expensive human annotation. In this paper, we present World to Code (W2C), a meticulously curated multi-modal data construction pipeline that organizes the final generation output into a Python code format. The pipeline leverages the VLM itself to extract cross-modal information via different prompts and filter the generated outputs again via a consistency filtering strategy. Experiments have demonstrated the high quality of W2C by improving various existing visual question answering and visual grounding benchmarks across different VLMs. Further analysis also demonstrates that the new code parsing ability of VLMs presents better cross-modal equivalence than the commonly used detail caption ability. Our code is available at https://github.com/foundation-multimodal-models/World2Code.
Remote-sensing applications often run on edge hardware that cannot host today's 7B-parameter multimodal language models. This paper introduces TinyRS, the first 2B-parameter multimodal small language model (MSLM) optimized for remote sensing tasks, and TinyRS-R1, its reasoning-augmented variant. Built upon Qwen2-VL-2B, TinyRS is trained through a four-stage pipeline: pre-training on million satellite images, instruction tuning on visual instruction examples, fine-tuning with Chain-of-Thought (CoT) annotations from the proposed reasoning dataset, and alignment via Group Relative Policy Optimization (GRPO). TinyRS-R1 achieves or surpasses the performance of recent 7B-parameter remote sensing models across classification, VQA, visual grounding, and open-ended question answering-while requiring just one-third of the memory and latency. Our analysis shows that CoT reasoning substantially benefits spatial grounding and scene understanding, while the non-reasoning TinyRS excels in concise, latency-sensitive VQA tasks. TinyRS-R1 represents the first domain-specialized MSLM with GRPO-aligned CoT reasoning for general-purpose remote sensing.
위장관(GI) 내시경 이미지 분석은 모델이 이미지에 대한 자유 형식의 임상적 질문에 답변해야 하는 시각적 질문 답변(VQA)으로 단일 레이블 분류에서 전환되고 있습니다. 최근의 비전-언어 모델(VLM)이 이 작업에서 유망한 답변 정확도를 달성하고 있지만, 임상적 도입을 위해서는 모델의 내부 표현이 답변의 근거가 되는 시각적 증거를 반영해야 합니다. 본 연구에서는 최소한의 추가 어노테이션으로 기존 VQA 데이터셋에서 보조 그라운딩(grounding) 및 설명 작업을 구성하는 간단한 다중 작업 미세조정 레시피를 제안합니다. 전문가가 주석을 추가한 용종 마스크를 직접 재사용하는 한편, Grad-CAM 위치 파악 기능이 있는 GI 도메인 사전학습 분류기를 통해 정답 마스크가 부족한 소견 범주에 대한 약한 지도(weak supervision)를 제공합니다. 세 가지 소형 VLM 백본을 Kvasir-VQA-x1에서 일치된 VQA 전용 및 다중 작업 레시피 하에 저순위 적응(LoRA)으로 미세조정하였으며, 분포 내(in-distribution) 및 분포 외(out-of-distribution) 데이터 모두에서 평가했을 때 답변 토큰과 관련 이미지 영역 간의 향상된 암시적 정렬과 함께 일관된 정확도 향상을 보임을 제시합니다.