PyTorchKR Papers

최신 논문

총 2,558편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2502.03692공식 코드 · Papers with Code

DocMIA: Document-Level Membership Inference Attacks against DocVQA Models

Document Visual Question Answering (DocVQA) has introduced a new paradigm for end-to-end document understanding, and quickly became one of the standard benchmarks for multimodal LLMs. Automating document processing workflows, driven by DocVQA models, presents significant potential for many business sectors. However, documents tend to contain highly sensitive information, raising concerns about privacy risks associated with training such DocVQA models. One significant privacy vulnerability, exploited by the membership inference attack, is the possibility for an adversary to determine if a particular record was part of the model's training data. In this paper, we introduce two novel membership inference attacks tailored specifically to DocVQA models. These attacks are designed for two different adversarial scenarios: a white-box setting, where the attacker has full access to the model architecture and parameters, and a black-box setting, where only the model's outputs are available. Notably, our attacks assume the adversary lacks access to auxiliary datasets, which is more realistic in practice but also more challenging. Our unsupervised methods outperform existing state-of-the-art membership inference attacks across a variety of DocVQA models and datasets, demonstrating their effectiveness and highlighting the privacy risks in this domain.

arXiv:2502.01576공식 코드 · Papers with Code

Robust-LLaVA: On the Effectiveness of Large-Scale Robust Image Encoders for Multi-modal Large Language Models

Multi-modal Large Language Models (MLLMs) excel in vision-language tasks but remain vulnerable to visual adversarial perturbations that can induce hallucinations, manipulate responses, or bypass safety mechanisms. Existing methods seek to mitigate these risks by applying constrained adversarial fine-tuning to CLIP vision encoders on ImageNet-scale data, ensuring their generalization ability is preserved. However, this limited adversarial training restricts robustness and broader generalization. In this work, we explore an alternative approach of leveraging existing vision classification models that have been adversarially pre-trained on large-scale data. Our analysis reveals two principal contributions: (1) the extensive scale and diversity of adversarial pre-training enables these models to demonstrate superior robustness against diverse adversarial threats, ranging from imperceptible perturbations to advanced jailbreaking attempts, without requiring additional adversarial training, and (2) end-to-end MLLM integration with these robust models facilitates enhanced adaptation of language components to robust visual features, outperforming existing plug-and-play methodologies on complex reasoning tasks. Through systematic evaluation across visual question-answering, image captioning, and jail-break attacks, we demonstrate that MLLMs trained with these robust models achieve superior adversarial robustness while maintaining favorable clean performance. Our framework achieves 2x and 1.5x average robustness gains in captioning and VQA tasks, respectively, and delivers over 10% improvement against jailbreak attacks. Code and pretrained models will be available at https://github.com/HashmatShadab/Robust-LLaVA.

arXiv:2502.00954

Hypo3D: Exploring Hypothetical Reasoning in 3D

The rise of vision-language foundation models marks an advancement in bridging the gap between human and machine capabilities in 3D scene reasoning. Existing 3D reasoning benchmarks assume real-time scene accessibility, which is impractical due to the high cost of frequent scene updates. To this end, we introduce Hypothetical 3D Reasoning, namely Hypo3D, a benchmark designed to evaluate models' ability to reason without access to real-time scene data. Models need to imagine the scene state based on a provided change description before reasoning. Hypo3D is formulated as a 3D Visual Question Answering (VQA) benchmark, comprising 7,727 context changes across 700 indoor scenes, resulting in 14,885 question-answer pairs. An anchor-based world frame is established for all scenes, ensuring consistent reference to a global frame for directional terms in context changes and QAs. Extensive experiments show that state-of-the-art foundation models struggle to reason in hypothetically changed scenes. This reveals a substantial performance gap compared to humans, particularly in scenarios involving movement changes and directional reasoning. Even when the context change is irrelevant to the question, models often incorrectly adjust their answers.

arXiv:2501.18011

Anatomy Might Be All You Need: Forecasting What to Do During Surgery

Surgical guidance can be delivered in various ways. In neurosurgery, spatial guidance and orientation are predominantly achieved through neuronavigation systems that reference pre-operative MRI scans. Recently, there has been growing interest in providing live guidance by analyzing video feeds from tools such as endoscopes. Existing approaches, including anatomy detection, orientation feedback, phase recognition, and visual question-answering, primarily focus on aiding surgeons in assessing the current surgical scene. This work aims to provide guidance on a finer scale, aiming to provide guidance by forecasting the trajectory of the surgical instrument, essentially addressing the question of what to do next. To address this task, we propose a model that not only leverages the historical locations of surgical instruments but also integrates anatomical features. Importantly, our work does not rely on explicit ground truth labels for instrument trajectories. Instead, the ground truth is generated by a detection model trained to detect both anatomical structures and instruments within surgical videos of a comprehensive dataset containing pituitary surgery videos. By analyzing the interaction between anatomy and instrument movements in these videos and forecasting future instrument movements, we show that anatomical features are a valuable asset in addressing this challenging task. To the best of our knowledge, this work is the first attempt to address this task for manually operated surgeries.

arXiv:2501.17811공식 코드 · Papers with Code

Janus-Pro: Unified Multimodal Understanding and Generation with Data and Model Scaling

In this work, we introduce Janus-Pro, an advanced version of the previous work Janus. Specifically, Janus-Pro incorporates (1) an optimized training strategy, (2) expanded training data, and (3) scaling to larger model size. With these improvements, Janus-Pro achieves significant advancements in both multimodal understanding and text-to-image instruction-following capabilities, while also enhancing the stability of text-to-image generation. We hope this work will inspire further exploration in the field. Code and models are publicly available.

arXiv:2501.15370

Scaling Large Vision-Language Models for Enhanced Multimodal Comprehension In Biomedical Image Analysis

Large language models (LLMs) have demonstrated immense capabilities in understanding textual data and are increasingly being adopted to help researchers accelerate scientific discovery through knowledge extraction (information retrieval), knowledge distillation (summarizing key findings and methodologies into concise forms), and knowledge synthesis (aggregating information from multiple scientific sources to address complex queries, generate hypothesis and formulate experimental plans). However, scientific data often exists in both visual and textual modalities. Vision language models (VLMs) address this by incorporating a pretrained vision backbone for processing images and a cross-modal projector that adapts image tokens into the LLM dimensional space, thereby providing richer multimodal comprehension. Nevertheless, off-the-shelf VLMs show limited capabilities in handling domain-specific data and are prone to hallucinations. We developed intelligent assistants finetuned from LLaVA models to enhance multimodal understanding in low-dose radiation therapy (LDRT)-a benign approach used in the treatment of cancer-related illnesses. Using multilingual data from 42,673 articles, we devise complex reasoning and detailed description tasks for visual question answering (VQA) benchmarks. Our assistants, trained on 50,882 image-text pairs, demonstrate superior performance over base models as evaluated using LLM-as-a-judge approach, particularly in reducing hallucination and improving domain-specific comprehension.

arXiv:2501.15140공식 코드 · Papers with Code

Analyzing and Boosting the Power of Fine-Grained Visual Recognition for Multi-modal Large Language Models

Multi-modal large language models (MLLMs) have shown remarkable abilities in various visual understanding tasks. However, MLLMs still struggle with fine-grained visual recognition (FGVR), which aims to identify subordinate-level categories from images. This can negatively impact more advanced capabilities of MLLMs, such as object-centric visual question answering and reasoning. In our study, we revisit three quintessential capabilities of MLLMs for FGVR, including object information extraction, category knowledge reserve, object-category alignment, and position of the root cause as a misalignment problem. To address this issue, we present Finedefics, an MLLM that enhances the model's FGVR capability by incorporating informative attribute descriptions of objects into the training phase. We employ contrastive learning on object-attribute pairs and attribute-category pairs simultaneously and use examples from similar but incorrect categories as hard negatives, naturally bringing representations of visual objects and category names closer. Extensive evaluations across multiple popular FGVR datasets demonstrate that Finedefics outperforms existing MLLMs of comparable parameter sizes, showcasing its remarkable efficacy. The code is available at https://github.com/PKU-ICST-MIPL/Finedefics_ICLR2025.

arXiv:2501.16300공식 코드 · Papers with Code

Large Models in Dialogue for Active Perception and Anomaly Detection

Autonomous aerial monitoring is an important task aimed at gathering information from areas that may not be easily accessible by humans. At the same time, this task often requires recognizing anomalies from a significant distance or not previously encountered in the past. In this paper, we propose a novel framework that leverages the advanced capabilities provided by Large Language Models (LLMs) to actively collect information and perform anomaly detection in novel scenes. To this end, we propose an LLM based model dialogue approach, in which two deep learning models engage in a dialogue to actively control a drone to increase perception and anomaly detection accuracy. We conduct our experiments in a high fidelity simulation environment where an LLM is provided with a predetermined set of natural language movement commands mapped into executable code functions. Additionally, we deploy a multimodal Visual Question Answering (VQA) model charged with the task of visual question answering and captioning. By engaging the two models in conversation, the LLM asks exploratory questions while simultaneously flying a drone into different parts of the scene, providing a novel way to implement active perception. By leveraging LLMs reasoning ability, we output an improved detailed description of the scene going beyond existing static perception approaches. In addition to information gathering, our approach is utilized for anomaly detection and our results demonstrate the proposed methods effectiveness in informing and alerting about potential hazards.

arXiv:2501.14520

Scene Understanding Enabled Semantic Communication with Open Channel Coding

As communication systems transition from symbol transmission to conveying meaningful information, sixth-generation (6G) networks emphasize semantic communication. This approach prioritizes high-level semantic information, improving robustness and reducing redundancy across modalities like text, speech, and images. However, traditional semantic communication faces limitations, including static coding strategies, poor generalization, and reliance on task-specific knowledge bases that hinder adaptability. To overcome these challenges, we propose a novel system combining scene understanding, Large Language Models (LLMs), and open channel coding, named \textbf{OpenSC}. Traditional systems rely on fixed domain-specific knowledge bases, limiting their ability to generalize. Our open channel coding approach leverages shared, publicly available knowledge, enabling flexible, adaptive encoding. This dynamic system reduces reliance on static task-specific data, enhancing adaptability across diverse tasks and environments. Additionally, we use scene graphs for structured semantic encoding, capturing object relationships and context to improve tasks like Visual Question Answering (VQA). Our approach selectively encodes key semantic elements, minimizing redundancy and improving transmission efficiency. Experimental results show significant improvements in both semantic understanding and efficiency, advancing the potential of adaptive, generalizable semantic communication in 6G networks.

arXiv:2501.12751공식 코드 · Papers with Code

Patent Figure Classification using Large Vision-language Models

Patent figure classification facilitates faceted search in patent retrieval systems, enabling efficient prior art search. Existing approaches have explored patent figure classification for only a single aspect and for aspects with a limited number of concepts. In recent years, large vision-language models (LVLMs) have shown tremendous performance across numerous computer vision downstream tasks, however, they remain unexplored for patent figure classification. Our work explores the efficacy of LVLMs in patent figure visual question answering (VQA) and classification, focusing on zero-shot and few-shot learning scenarios. For this purpose, we introduce new datasets, PatFigVQA and PatFigCLS, for fine-tuning and evaluation regarding multiple aspects of patent figures~(i.e., type, projection, patent class, and objects). For a computational-effective handling of a large number of classes using LVLM, we propose a novel tournament-style classification strategy that leverages a series of multiple-choice questions. Experimental results and comparisons of multiple classification approaches based on LVLMs and Convolutional Neural Networks (CNNs) in few-shot settings show the feasibility of the proposed approaches.

arXiv:2501.12697

Combining Knowledge Graph and LLMs for Enhanced Zero-shot Visual Question Answering

Zero-shot visual question answering (ZS-VQA), an emerged critical research area, intends to answer visual questions without providing training samples. Existing research in ZS-VQA has proposed to leverage knowledge graphs or large language models (LLMs), respectively, as external information sources to help VQA model comprehend images and questions. However, LLMs often struggle in accurately interpreting specific question meanings. Meanwhile, although knowledge graph has rich entity relationships, it is challenging to effectively connect entities to individual image content for visual question answers. In this paper, we propose a novel design to combine knowledge graph and LLMs for zero-shot visual question answer. Our approach uses LLMs' powerful understanding capabilities to accurately interpret image content through a strategic question search mechanism. Meanwhile, the knowledge graph is used to expand and connect users' queries to the image content for better visual question answering. An optimization algorithm is further used to determine the optimal weights for the loss functions derived from different information sources, towards a globally optimal set of candidate answers. Experimental results on two benchmark datasets demonstrate that our model achieves state-of-the-art (SOTA) performance. Both source code and benchmark data will be released for public access.

arXiv:2501.10674

Can Multimodal LLMs do Visual Temporal Understanding and Reasoning? The answer is No!

Multimodal Large Language Models (MLLMs) have achieved significant advancements in tasks like Visual Question Answering (VQA) by leveraging foundational Large Language Models (LLMs). However, their abilities in specific areas such as temporal understanding, which is crucial for comprehending real-world dynamics, remain underexplored. To address this, we propose a challenging evaluation benchmark named TemporalVQA, consisting of two parts: (1) Temporal Order Understanding and (2) Time-lapse Estimation. The first part requires MLLMs to determine the sequence of events by analyzing temporally consecutive video frames. The second part presents image pairs with varying time differences, framed as multiple-choice questions, asking MLLMs to estimate the time-lapse between images with options ranging from seconds to years. Our evaluations of advanced MLLMs, including models like GPT-4o and Gemini-1.5-Pro, reveal significant challenges: GPT-4o achieved only 43.8% average consistent accuracy in temporal order tasks and 70% in time-lapse estimation, with open-source models performing even less effectively. These findings underscore the limitations of current MLLMs in visual temporal understanding and reasoning, highlighting the need for further improvements in their temporal capabilities. Our dataset can be found at https://huggingface.co/datasets/fazliimam/temporal-vqa.

arXiv:2501.12327공식 코드 · Papers with Code

VARGPT: Unified Understanding and Generation in a Visual Autoregressive Multimodal Large Language Model

We present VARGPT, a novel multimodal large language model (MLLM) that unifies visual understanding and generation within a single autoregressive framework. VARGPT employs a next-token prediction paradigm for visual understanding and a next-scale prediction paradigm for visual autoregressive generation. VARGPT innovatively extends the LLaVA architecture, achieving efficient scale-wise autoregressive visual generation within MLLMs while seamlessly accommodating mixed-modal input and output within a single model framework. Our VARGPT undergoes a three-stage unified training process on specially curated datasets, comprising a pre-training phase and two mixed visual instruction-tuning phases. The unified training strategy are designed to achieve alignment between visual and textual features, enhance instruction following for both understanding and generation, and improve visual generation quality, respectively. Despite its LLAVA-based architecture for multimodel understanding, VARGPT significantly outperforms LLaVA-1.5 across various vision-centric benchmarks, such as visual question-answering and reasoning tasks. Notably, VARGPT naturally supports capabilities in autoregressive visual generation and instruction-to-image synthesis, showcasing its versatility in both visual understanding and generation tasks. Project page is at: \url{https://vargpt-1.github.io/}

arXiv:2501.06189

A Multimodal Social Agent

In recent years, large language models (LLMs) have demonstrated remarkable progress in common-sense reasoning tasks. This ability is fundamental to understanding social dynamics, interactions, and communication. However, the potential of integrating computers with these social capabilities is still relatively unexplored. However, the potential of integrating computers with these social capabilities is still relatively unexplored. This paper introduces MuSA, a multimodal LLM-based agent that analyzes text-rich social content tailored to address selected human-centric content analysis tasks, such as question answering, visual question answering, title generation, and categorization. It uses planning, reasoning, acting, optimizing, criticizing, and refining strategies to complete a task. Our approach demonstrates that MuSA can automate and improve social content analysis, helping decision-making processes across various applications. We have evaluated our agent's capabilities in question answering, title generation, and content categorization tasks. MuSA performs substantially better than our baselines.

arXiv:2501.09446

Double Visual Defense: Adversarial Pre-training and Instruction Tuning for Improving Vision-Language Model Robustness

This paper investigates the robustness of vision-language models against adversarial visual perturbations and introduces a novel ``double visual defense" to enhance this robustness. Unlike previous approaches that resort to lightweight adversarial fine-tuning of a pre-trained CLIP model, we perform large-scale adversarial vision-language pre-training from scratch using web-scale data. We then strengthen the defense by incorporating adversarial visual instruction tuning. The resulting models from each stage, $\Delta$CLIP and $\Delta^2$LLaVA, show substantially enhanced zero-shot robustness and set a new state-of-the-art in adversarial defense for vision-language models. For example, the adversarial robustness of $\Delta$CLIP surpasses that of the previous best models on ImageNet-1k by ~20%. %For example, $\Delta$CLIP surpasses the previous best models on ImageNet-1k by ~20% in terms of adversarial robustness. Similarly, compared to prior art, $\Delta^2$LLaVA brings a ~30% robustness improvement to image captioning task and a ~20% robustness improvement to visual question answering task. Furthermore, our models exhibit stronger zero-shot recognition capability, fewer hallucinations, and superior reasoning performance compared to baselines. Our project page is https://doublevisualdefense.github.io/.

arXiv:2501.09167

Embodied Scene Understanding for Vision Language Models via MetaVQA

Vision Language Models (VLMs) demonstrate significant potential as embodied AI agents for various mobility applications. However, a standardized, closed-loop benchmark for evaluating their spatial reasoning and sequential decision-making capabilities is lacking. To address this, we present MetaVQA: a comprehensive benchmark designed to assess and enhance VLMs' understanding of spatial relationships and scene dynamics through Visual Question Answering (VQA) and closed-loop simulations. MetaVQA leverages Set-of-Mark prompting and top-down view ground-truth annotations from nuScenes and Waymo datasets to automatically generate extensive question-answer pairs based on diverse real-world traffic scenarios, ensuring object-centric and context-rich instructions. Our experiments show that fine-tuning VLMs with the MetaVQA dataset significantly improves their spatial reasoning and embodied scene comprehension in safety-critical simulations, evident not only in improved VQA accuracies but also in emerging safety-aware driving maneuvers. In addition, the learning demonstrates strong transferability from simulation to real-world observation. Code and data will be publicly available at https://metadriverse.github.io/metavqa .

arXiv:2501.09720공식 코드 · Papers with Code

A Simple Aerial Detection Baseline of Multimodal Language Models

The multimodal language models (MLMs) based on generative pre-trained Transformer are considered powerful candidates for unifying various domains and tasks. MLMs developed for remote sensing (RS) have demonstrated outstanding performance in multiple tasks, such as visual question answering and visual grounding. In addition to visual grounding that detects specific objects corresponded to given instruction, aerial detection, which detects all objects of multiple categories, is also a valuable and challenging task for RS foundation models. However, aerial detection has not been explored by existing RS MLMs because the autoregressive prediction mechanism of MLMs differs significantly from the detection outputs. In this paper, we present a simple baseline for applying MLMs to aerial detection for the first time, named LMMRotate. Specifically, we first introduce a normalization method to transform detection outputs into textual outputs to be compatible with the MLM framework. Then, we propose a evaluation method, which ensures a fair comparison between MLMs and conventional object detection models. We construct the baseline by fine-tuning open-source general-purpose MLMs and achieve impressive detection performance comparable to conventional detector. We hope that this baseline will serve as a reference for future MLM development, enabling more comprehensive capabilities for understanding RS images. Code is available at https://github.com/Li-Qingyun/mllm-mmrotate.

arXiv:2501.08597

Dynamic Knowledge Integration for Enhanced Vision-Language Reasoning

Large Vision-Language Models (LVLMs) have demonstrated impressive capabilities in multimodal tasks, but their performance is often constrained by the lack of external knowledge integration, limiting their ability to handle knowledge-intensive tasks such as visual question answering and reasoning. To address this challenge, we propose a novel method, Adaptive Knowledge-Guided Pretraining for Large Vision-Language Models (AKGP-LVLM), which dynamically incorporates structured and unstructured knowledge into LVLMs during pretraining and fine-tuning. Our approach employs a knowledge encoder to represent external knowledge, a retrieval mechanism to select task-relevant information, and a dynamic adaptor to align multimodal and knowledge representations effectively. We evaluate our method on four benchmark datasets, demonstrating significant performance improvements over state-of-the-art models. Furthermore, human evaluations highlight the superior correctness and relevance of our model's outputs. Extensive analyses confirm the robustness, efficiency, and scalability of AKGP-LVLM, making it a compelling solution for real-world knowledge-intensive tasks.

arXiv:2501.08131

SAR Strikes Back: A New Hope for RSVQA

Remote sensing visual question answering (RSVQA) is a task that automatically extracts information from satellite images and processes a question to predict the answer from the images in textual form, helping with the interpretation of the image. While different methods have been proposed to extract information from optical images with different spectral bands and resolutions, no method has been proposed to answer questions from Synthetic Aperture Radar (SAR) images. SAR images capture electromagnetic information from the scene, and are less affected by atmospheric conditions, such as clouds. In this work, our objective is to introduce SAR in the RSVQA task, finding the best way to use this modality. In our research, we carry out a study on different pipelines for the task of RSVQA taking into account information from both SAR and optical data. To this purpose, we also present a dataset that allows for the introduction of SAR images in the RSVQA framework. We propose two different models to include the SAR modality. The first one is an end-to-end method in which we add an additional encoder for the SAR modality. In the second approach, we build on a two-stage framework. First, relevant information is extracted from SAR and, optionally, optical data. This information is then translated into natural language to be used in the second step which only relies on a language model to provide the answer. We find that the second pipeline allows us to obtain good results with SAR images alone. We then try various types of fusion methods to use SAR and optical images together, finding that a fusion at the decision level achieves the best results on the proposed dataset. We show that SAR data offers additional information when fused with the optical modality, particularly for questions related to specific land cover classes, such as water areas.

arXiv:2501.07783공식 코드 · Papers with Code

Parameter-Inverted Image Pyramid Networks for Visual Perception and Multimodal Understanding

Image pyramids are widely adopted in top-performing methods to obtain multi-scale features for precise visual perception and understanding. However, current image pyramids use the same large-scale model to process multiple resolutions of images, leading to significant computational cost. To address this challenge, we propose a novel network architecture, called Parameter-Inverted Image Pyramid Networks (PIIP). Specifically, PIIP uses pretrained models (ViTs or CNNs) as branches to process multi-scale images, where images of higher resolutions are processed by smaller network branches to balance computational cost and performance. To integrate information from different spatial scales, we further propose a novel cross-branch feature interaction mechanism. To validate PIIP, we apply it to various perception models and a representative multimodal large language model called LLaVA, and conduct extensive experiments on various tasks such as object detection, segmentation, image classification and multimodal understanding. PIIP achieves superior performance compared to single-branch and existing multi-resolution approaches with lower computational cost. When applied to InternViT-6B, a large-scale vision foundation model, PIIP can improve its performance by 1%-2% on detection and segmentation with only 40%-60% of the original computation, finally achieving 60.0 box AP on MS COCO and 59.7 mIoU on ADE20K. For multimodal understanding, our PIIP-LLaVA achieves 73.0% accuracy on TextVQA and 74.5% on MMBench with only 2.8M training data. Our code is released at https://github.com/OpenGVLab/PIIP.