Automated radiology report generation aims to expedite the tedious and error-prone reporting process for radiologists. While recent works have made progress, learning to align medical images and textual findings remains challenging due to the relative scarcity of labeled medical data. For example, datasets for this task are much smaller than those used for image captioning in computer vision. In this work, we propose to transfer representations from CLIP, a large-scale pre-trained vision-language model, to better capture cross-modal semantics between images and texts. However, directly applying CLIP is suboptimal due to the domain gap between natural images and radiology. To enable efficient adaptation, we introduce UniCrossAdapter, lightweight adapter modules that are incorporated into CLIP and fine-tuned on the target task while keeping base parameters fixed. The adapters are distributed across modalities and their interaction to enhance vision-language alignment. Experiments on two public datasets demonstrate the effectiveness of our approach, advancing state-of-the-art in radiology report generation. The proposed transfer learning framework provides a means of harnessing semantic knowledge from large-scale pre-trained models to tackle data-scarce medical vision-language tasks. Code is available at https://github.com/chauncey-tow/MRG-CLIP.
In the rapidly evolving field of artificial intelligence, multimodal models, e.g., integrating vision and language into visual-language models (VLMs), have become pivotal for many applications, ranging from image captioning to multimodal search engines. Among these models, the Contrastive Language-Image Pre-training (CLIP) model has demonstrated remarkable performance in understanding and generating nuanced relationships between text and images. However, the conventional training of such models often requires centralized aggregation of vast datasets, posing significant privacy and data governance challenges. To address these concerns, this paper proposes a novel approach that leverages Federated Learning and parameter-efficient adapters, i.e., Low-Rank Adaptation (LoRA), to train VLMs. This methodology preserves data privacy by training models across decentralized data sources and ensures model adaptability and efficiency through LoRA's parameter-efficient fine-tuning. Our approach accelerates training time by up to 34.72 times and requires 2.47 times less memory usage than full fine-tuning.
Multimodal systems have highly complex processing pipelines and are pretrained over large datasets before being fine-tuned for specific tasks such as visual captioning. However, it becomes hard to disentangle what the model learns during the fine-tuning process from what it already knows due to its pretraining. In this work, we learn a probabilistic model using Hybrid Markov Logic Networks (HMLNs) over the training examples by relating symbolic knowledge (extracted from the caption) with visual features (extracted from the image). For a generated caption, we quantify the influence of training examples based on the HMLN distribution using probabilistic inference. We evaluate two types of inference procedures on the MSCOCO dataset for different types of captioning models. Our results show that for BLIP2 (a model that uses a LLM), the fine-tuning may have smaller influence on the knowledge the model has acquired since it may have more general knowledge to perform visual captioning as compared to models that do not use a LLM
Recent progress in (multimodal) large language models ((M)LLMs) has shifted focus from pre-training to inference-time compute scaling and post-training optimization, driven by concerns over limited high-quality real-world data. However, these strategies alone are insufficient for advancing model capabilities. We hypothesize that effective model improvement requires a strong synergy among pre-training, inference-time compute scaling, and post-training optimization. In this paper, we validate this hypothesis in the context of multimodal pre-training for foundation MLLM construction. We introduce Self-Improving cognition (SIcog), a self-learning framework for constructing next-generation foundation MLLMs by imparting multimodal knowledge and enhancing their systematic cognitive capabilities through multimodal pre-training with self-generated data. Specifically, we introduce Chain-of-Description, a step-by-step visual understanding method to improve comprehensive perception, and integrate structured chain-of-thought (CoT) reasoning to support in-depth multimodal reasoning. SIcog first equips a base model with systematic perception and reasoning using minimal external supervision. The enhanced model then generates candidate image captions and CoT-style reasoning responses for unlabeled images and image-question pairs across diverse tasks, which are curated through a self-consistency mechanism. These curated samples are subsequently used for large-scale multimodal pre-training, completing a self-learning cycle that strengthens the model's cognitive foundation. Extensive experiments demonstrate that SIcog produces next-generation foundation MLLMs with substantially improved multimodal cognition, outperforming prevailing pre-training approaches. These findings empirically establish SIcog as a promising framework for realizing a complete self-improving paradigm.
This paper introduces a holistic vision-language foundation model tailored for remote sensing, named Falcon. Falcon offers a unified, prompt-based paradigm that effectively executes comprehensive and complex remote sensing tasks. Falcon demonstrates powerful understanding and reasoning abilities at the image, region, and pixel levels. Specifically, given simple natural language instructions and remote sensing images, Falcon can produce impressive results in text form across 14 distinct tasks, i.e., image classification, object detection, segmentation, image captioning, and etc. To facilitate Falcon's training and empower its representation capacity to encode rich spatial and semantic information, we developed Falcon_SFT, a large-scale, multi-task, instruction-tuning dataset in the field of remote sensing. The Falcon_SFT dataset consists of approximately 78 million high-quality data samples, covering 5.6 million multi-spatial resolution and multi-view remote sensing images with diverse instructions. It features hierarchical annotations and undergoes manual sampling verification to ensure high data quality and reliability. Extensive comparative experiments are conducted, which verify that Falcon achieves remarkable performance over 67 datasets and 14 tasks, despite having only 0.7B parameters. We release the complete dataset, code, and model weights at https://github.com/TianHuiLab/Falcon, hoping to help further develop the open-source community.
Writing Assistants (e.g., Grammarly, Microsoft Copilot) traditionally generate diverse image captions by employing syntactic and semantic variations to describe image components. However, human-written captions prioritize conveying a central message alongside visual descriptions using pragmatic cues. To enhance pragmatic diversity, it is essential to explore alternative ways of communicating these messages in conjunction with visual content. To address this challenge, we propose RONA, a novel prompting strategy for Multi-modal Large Language Models (MLLM) that leverages Coherence Relations as an axis for variation. We demonstrate that RONA generates captions with better overall diversity and ground-truth alignment, compared to MLLM baselines across multiple domains. Our code is available at: https://github.com/aashish2000/RONA
In the context of pressing climate change challenges and the significant biodiversity loss among arthropods, automated taxonomic classification from organismal images is a subject of intense research. However, traditional AI pipelines based on deep neural visual architectures such as CNNs or ViTs face limitations such as degraded performance on the long-tail of classes and the inability to reason about their predictions. We integrate image captioning and retrieval-augmented generation (RAG) with large language models (LLMs) to enhance biodiversity monitoring, showing particular promise for characterizing rare and unknown arthropod species. While a naive Vision-Language Model (VLM) excels in classifying images of common species, the RAG model enables classification of rarer taxa by matching explicit textual descriptions of taxonomic features to contextual biodiversity text data from external sources. The RAG model shows promise in reducing overconfidence and enhancing accuracy relative to naive LLMs, suggesting its viability in capturing the nuances of taxonomic hierarchy, particularly at the challenging family and genus levels. Our findings highlight the potential for modern vision-language AI pipelines to support biodiversity conservation initiatives, emphasizing the role of comprehensive data curation and collaboration with citizen science platforms to improve species identification, unknown species characterization and ultimately inform conservation strategies.
Cross-lingual transfer enables vision-language models (VLMs) to perform vision tasks in various languages with training data only in one language. Current approaches rely on large pre-trained multilingual language models. However, they face the curse of multilinguality, sacrificing downstream task performance for multilingual capabilities, struggling with lexical ambiguities, and falling behind recent advances. In this work, we study the scaling laws of systematic generalization with monolingual VLMs for multilingual tasks, focusing on the impact of model size and seen training samples. We propose Florenz, a monolingual encoder-decoder VLM with 0.4B to 11.2B parameters combining the pre-trained VLM Florence-2 and the large language model Gemma-2. Florenz is trained with varying compute budgets on a synthetic dataset that features intentionally incomplete language coverage for image captioning, thus, testing generalization from the fully covered translation task. We show that not only does indirectly learning unseen task-language pairs adhere to a scaling law, but also that with our data generation pipeline and the proposed Florenz model family, image captioning abilities can emerge in a specific language even when only data for the translation task is available. Fine-tuning on a mix of downstream datasets yields competitive performance and demonstrates promising scaling trends in multimodal machine translation (Multi30K, CoMMuTE), lexical disambiguation (CoMMuTE), and image captioning (Multi30K, XM3600, COCO Karpathy).
When we train models on biased ML datasets, they not only learn these biases but can inflate them at test time - a phenomenon called bias amplification. To measure bias amplification in ML datasets, many co-occurrence-based metrics have been proposed. Co-occurrence-based metrics are effective in measuring bias amplification in simple problems like image classification. However, these metrics are ineffective for complex problems like image captioning as they cannot capture the semantics of a caption. To measure bias amplification in captions, prior work introduced a predictability-based metric called Leakage in Captioning (LIC). While LIC captures the semantics and context of captions, it has limitations. LIC cannot identify the direction in which bias is amplified, poorly estimates dataset bias due to a weak vocabulary substitution strategy, and is highly sensitive to attacker models (a hyperparameter in predictability-based metrics). To overcome these issues, we propose Directional Predictability Amplification in Captioning (DPAC). DPAC measures directional bias amplification in captions, provides a better estimate of dataset bias using an improved substitution strategy, and is less sensitive to attacker models. Our experiments on the COCO captioning dataset show how DPAC is the most reliable metric to measure bias amplification in captions.
Vision-Language Models (VLMs) based on Mixture-of-Experts (MoE) architectures have emerged as a pivotal paradigm in multimodal understanding, offering a powerful framework for integrating visual and linguistic information. However, the increasing complexity and diversity of tasks present significant challenges in coordinating load balancing across heterogeneous visual experts, where optimizing one specialist's performance often compromises others' capabilities. To address task heterogeneity and expert load imbalance, we propose Astrea, a novel multi-expert collaborative VLM architecture based on progressive pre-alignment. Astrea introduces three key innovations: 1) A heterogeneous expert coordination mechanism that integrates four specialized models (detection, segmentation, classification, captioning) into a comprehensive expert matrix covering essential visual comprehension elements; 2) A dynamic knowledge fusion strategy featuring progressive pre-alignment to harmonize experts within the VLM latent space through contrastive learning, complemented by probabilistically activated stochastic residual connections to preserve knowledge continuity; 3) An enhanced optimization framework utilizing momentum contrastive learning for long-range dependency modeling and adaptive weight allocators for real-time expert contribution calibration. Extensive evaluations across 12 benchmark tasks spanning VQA, image captioning, and cross-modal retrieval demonstrate Astrea's superiority over state-of-the-art models, achieving an average performance gain of +4.7\%. This study provides the first empirical demonstration that progressive pre-alignment strategies enable VLMs to overcome task heterogeneity limitations, establishing new methodological foundations for developing general-purpose multimodal agents.
Remote Sensing Image Captioning (RSIC) is a cross-modal field bridging vision and language, aimed at automatically generating natural language descriptions of features and scenes in remote sensing imagery. Despite significant advances in developing sophisticated methods and large-scale datasets for training vision-language models (VLMs), two critical challenges persist: the scarcity of non-English descriptive datasets and the lack of multilingual capability evaluation for models. These limitations fundamentally impede the progress and practical deployment of RSIC, particularly in the era of large VLMs. To address these challenges, this paper presents several significant contributions to the field. First, we introduce and analyze BRSIC (Bilingual Remote Sensing Image Captioning), a comprehensive bilingual dataset that enriches three established English RSIC datasets with Chinese descriptions, encompassing 13,634 images paired with 68,170 bilingual captions. Building upon this foundation, we develop a systematic evaluation framework that addresses the prevalent inconsistency in evaluation protocols, enabling rigorous assessment of model performance through standardized retraining procedures on BRSIC. Furthermore, we present an extensive empirical study of eight state-of-the-art large vision-language models (LVLMs), examining their capabilities across multiple paradigms including zero-shot inference, supervised fine-tuning, and multi-lingual training. This comprehensive evaluation provides crucial insights into the strengths and limitations of current LVLMs in handling multilingual remote sensing tasks. Additionally, our cross-dataset transfer experiments reveal interesting findings. The code and data will be available at https://github.com/mrazhou/BRSIC.
In Recent Years, Digital Technologies Have Made Significant Strides In Augmenting-Human-Health, Cognition, And Perception, Particularly Within The Field Of Computational-Pathology. This Paper Presents A Novel Approach To Enhancing The Analysis Of Histopathology Images By Leveraging A Mult-modal-Model That Combines Vision Transformers (Vit) With Gpt-2 For Image Captioning. The Model Is Fine-Tuned On The Specialized Arch-Dataset, Which Includes Dense Image Captions Derived From Clinical And Academic Resources, To Capture The Complexities Of Pathology Images Such As Tissue Morphologies, Staining Variations, And Pathological Conditions. By Generating Accurate, Contextually Captions, The Model Augments The Cognitive Capabilities Of Healthcare Professionals, Enabling More Efficient Disease Classification, Segmentation, And Detection. The Model Enhances The Perception Of Subtle Pathological Features In Images That Might Otherwise Go Unnoticed, Thereby Improving Diagnostic Accuracy. Our Approach Demonstrates The Potential For Digital Technologies To Augment Human Cognitive Abilities In Medical Image Analysis, Providing Steps Toward More Personalized And Accurate Healthcare Outcomes.
This paper aims to address the challenge of hallucinations in Multimodal Large Language Models (MLLMs) particularly for dense image captioning tasks. To tackle the challenge, we identify the current lack of a metric that finely measures the caption quality in concept level. We hereby introduce HalFscore, a novel metric built upon the language graph and is designed to evaluate both the accuracy and completeness of dense captions at a granular level. Additionally, we identify the root cause of hallucination as the model's over-reliance on its language prior. To address this, we propose PerturboLLaVA, which reduces the model's reliance on the language prior by incorporating adversarially perturbed text during training. This method enhances the model's focus on visual inputs, effectively reducing hallucinations and producing accurate, image-grounded descriptions without incurring additional computational overhead. PerturboLLaVA significantly improves the fidelity of generated captions, outperforming existing approaches in handling multimodal hallucinations and achieving improved performance across general multimodal benchmarks.
Aligning large vision-language models (LVLMs) with human preferences is challenging due to the scarcity of fine-grained, high-quality, and multimodal preference data without human annotations. Existing methods relying on direct distillation often struggle with low-confidence data, leading to suboptimal performance. To address this, we propose CAREVL, a novel method for preference reward modeling by reliably using both high- and low-confidence data. First, a cluster of auxiliary expert models (textual reward models) innovatively leverages image captions as weak supervision signals to filter high-confidence data. The high-confidence data are then used to fine-tune the LVLM. Second, low-confidence data are used to generate diverse preference samples using the fine-tuned LVLM. These samples are then scored and selected to construct reliable chosen-rejected pairs for further training. CAREVL achieves performance improvements over traditional distillation-based methods on VL-RewardBench and MLLM-as-a-Judge benchmark, demonstrating its effectiveness. The code will be released soon.
Vision-Language Models (VLMs) have advanced multi-modal tasks like image captioning, visual question answering, and reasoning. However, they often generate hallucinated outputs inconsistent with the visual context or prompt, limiting reliability in critical applications like autonomous driving and medical imaging. Existing studies link hallucination to statistical biases, language priors, and biased feature learning but lack a structured causal understanding. In this work, we introduce a causal perspective to analyze and mitigate hallucination in VLMs. We hypothesize that hallucination arises from unintended direct influences of either the vision or text modality, bypassing proper multi-modal fusion. To address this, we construct a causal graph for VLMs and employ counterfactual analysis to estimate the Natural Direct Effect (NDE) of vision, text, and their cross-modal interaction on the output. We systematically identify and mitigate these unintended direct effects to ensure that responses are primarily driven by genuine multi-modal fusion. Our approach consists of three steps: (1) designing structural causal graphs to distinguish correct fusion pathways from spurious modality shortcuts, (2) estimating modality-specific and cross-modal NDE using perturbed image representations, hallucinated text embeddings, and degraded visual inputs, and (3) implementing a test-time intervention module to dynamically adjust the model's dependence on each modality. Experimental results demonstrate that our method significantly reduces hallucination while preserving task performance, providing a robust and interpretable framework for improving VLM reliability. To enhance accessibility and reproducibility, our code is publicly available at https://github.com/TREE985/Treble-Counterfactual-VLMs.
Most existing works in image caption synthesis use computation heavy deep neural networks and generates image descriptions in English language. This often restricts this important assistive tool for widespread use across language and accessibility barriers. This work presents AC-Lite, a computationally efficient model for image captioning in low-resource Assamese language. AC-Lite reduces computational requirements by replacing computation-heavy deep network components with lightweight alternatives. The AC-Lite model is designed through extensive ablation experiments with different image feature extractor networks and language decoders. A combination of ShuffleNetv2x1.5 with GRU based language decoder along with bilinear attention is found to provide the best performance with minimum compute. AC-Lite was observed to achieve an 82.3 CIDEr score on the COCO-AC dataset with 2.45 GFLOPs and 22.87M parameters.
Image captioning tasks usually use two-stage training to complete model optimization. The first stage uses cross-entropy as the loss function for optimization, and the second stage uses self-critical sequence training (SCST) for reinforcement learning optimization. However, the SCST algorithm has certain defects. SCST relies only on a single greedy decoding result as a baseline. If the model itself is not stable enough, the greedy decoding result may be relatively worst, which will lead to a high variance of advantage estimation, further leading to unstable policy updates. In addition, SCST only compares one sampling result with the greedy decoding result, and the generation diversity is limited, which may fall into a local optimum. In this paper, we propose using the latest Group Relative Policy Optimization (GRPO) reinforcement learning algorithm as an optimization solution for the second stage. GRPO generates multiple candidate captions for the input image and then continuously optimizes the model through intragroup comparison. By constraining the amplitude of policy updates and KL divergence, the stability of the model during training is greatly guaranteed. In addition, compared to SCST, which only samples one answer, GRPO samples and generates multiple answers. Multiple candidate answers in the group cover a wider solution space. Combined with KL divergence constraints, GRPO can improve diversity while ensuring model stability. The code for this article is available at https://github.com/liangxu-one/ms-models/tree/image_caption_grpo/research/arxiv_papers/Image_Caption_GRPO.
Existing Video Scene Graph Generation (VidSGG) studies are trained in a fully supervised manner, which requires all frames in a video to be annotated, thereby incurring high annotation cost compared to Image Scene Graph Generation (ImgSGG). Although the annotation cost of VidSGG can be alleviated by adopting a weakly supervised approach commonly used for ImgSGG (WS-ImgSGG) that uses image captions, there are two key reasons that hinder such a naive adoption: 1) Temporality within video captions, i.e., unlike image captions, video captions include temporal markers (e.g., before, while, then, after) that indicate time related details, and 2) Variability in action duration, i.e., unlike human actions in image captions, human actions in video captions unfold over varying duration. To address these issues, we propose a Natural Language-based Video Scene Graph Generation (NL-VSGG) framework that only utilizes the readily available video captions for training a VidSGG model. NL-VSGG consists of two key modules: Temporality-aware Caption Segmentation (TCS) module and Action Duration Variability-aware caption-frame alignment (ADV) module. Specifically, TCS segments the video captions into multiple sentences in a temporal order based on a Large Language Model (LLM), and ADV aligns each segmented sentence with appropriate frames considering the variability in action duration. Our approach leads to a significant enhancement in performance compared to simply applying the WS-ImgSGG pipeline to VidSGG on the Action Genome dataset. As a further benefit of utilizing the video captions as weak supervision, we show that the VidSGG model trained by NL-VSGG is able to predict a broader range of action classes that are not included in the training data, which makes our framework practical in reality.
Large Vision-Language Models (LVLMs) integrate image encoders with Large Language Models (LLMs) to process multi-modal inputs and perform complex visual tasks. However, they often generate hallucinations by describing non-existent objects or attributes, compromising their reliability. This study analyzes hallucination patterns in image captioning, showing that not all tokens in the generation process are influenced by image input and that image dependency can serve as a useful signal for hallucination detection. To address this, we develop an automated pipeline to identify hallucinated objects and train a token-level classifier using hidden representations from parallel inference passes-with and without image input. Leveraging this classifier, we introduce a decoding strategy that effectively controls hallucination rates in image captioning at inference time.
High-quality textual training data is essential for the success of multimodal data processing tasks, yet outputs from image captioning models like BLIP and GIT often contain errors and anomalies that are difficult to rectify using rule-based methods. While recent work addressing this issue has predominantly focused on using GPT models for data preprocessing on relatively simple public datasets, there is a need to explore a broader range of Large Language Models (LLMs) and tackle more challenging and diverse datasets. In this study, we investigate the use of multiple LLMs, including LLaMA 3.1 70B, GPT-4 Turbo, and Sonnet 3.5 v2, to refine and clean the textual outputs of BLIP and GIT. We assess the impact of LLM-assisted data cleaning by comparing downstream-task (SemEval 2024 Subtask "Multilabel Persuasion Detection in Memes") models trained on cleaned versus non-cleaned data. While our experimental results show improvements when using LLM-cleaned captions, statistical tests reveal that most of these improvements are not significant. This suggests that while LLMs have the potential to enhance data cleaning and repairing, their effectiveness may be limited depending on the context they are applied to, the complexity of the task, and the level of noise in the text. Our findings highlight the need for further research into the capabilities and limitations of LLMs in data preprocessing pipelines, especially when dealing with challenging datasets, contributing empirical evidence to the ongoing discussion about integrating LLMs into data preprocessing pipelines.