PyTorchKR Papers

최신 논문

총 1,759편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2408.01959

Dataset Scale and Societal Consistency Mediate Facial Impression Bias in Vision-Language AI

Multimodal AI models capable of associating images and text hold promise for numerous domains, ranging from automated image captioning to accessibility applications for blind and low-vision users. However, uncertainty about bias has in some cases limited their adoption and availability. In the present work, we study 43 CLIP vision-language models to determine whether they learn human-like facial impression biases, and we find evidence that such biases are reflected across three distinct CLIP model families. We show for the first time that the the degree to which a bias is shared across a society predicts the degree to which it is reflected in a CLIP model. Human-like impressions of visually unobservable attributes, like trustworthiness and sexuality, emerge only in models trained on the largest dataset, indicating that a better fit to uncurated cultural data results in the reproduction of increasingly subtle social biases. Moreover, we use a hierarchical clustering approach to show that dataset size predicts the extent to which the underlying structure of facial impression bias resembles that of facial impression bias in humans. Finally, we show that Stable Diffusion models employing CLIP as a text encoder learn facial impression biases, and that these biases intersect with racial biases in Stable Diffusion XL-Turbo. While pretrained CLIP models may prove useful for scientific studies of bias, they will also require significant dataset curation when intended for use as general-purpose models in a zero-shot setting.

arXiv:2408.02595

Modelling Visual Semantics via Image Captioning to extract Enhanced Multi-Level Cross-Modal Semantic Incongruity Representation with Attention for Multimodal Sarcasm Detection

Sarcasm is a type of irony, characterized by an inherent mismatch between the literal interpretation and the intended connotation. Though sarcasm detection in text has been extensively studied, there are situations in which textual input alone might be insufficient to perceive sarcasm. The inclusion of additional contextual cues, such as images, is essential to recognize sarcasm in social media data effectively. This study presents a novel framework for multimodal sarcasm detection that can process input triplets. Two components of these triplets comprise the input text and its associated image, as provided in the datasets. Additionally, a supplementary modality is introduced in the form of descriptive image captions. The motivation behind incorporating this visual semantic representation is to more accurately capture the discrepancies between the textual and visual content, which are fundamental to the sarcasm detection task. The primary contributions of this study are: (1) a robust textual feature extraction branch that utilizes a cross-lingual language model; (2) a visual feature extraction branch that incorporates a self-regulated residual ConvNet integrated with a lightweight spatially aware attention module; (3) an additional modality in the form of image captions generated using an encoder-decoder architecture capable of reading text embedded in images; (4) distinct attention modules to effectively identify the incongruities between the text and two levels of image representations; (5) multi-level cross-domain semantic incongruity representation achieved through feature fusion. Compared with cutting-edge baselines, the proposed model achieves the best accuracy of 92.89% and 64.48%, respectively, on the Twitter multimodal sarcasm and MultiBully datasets.

arXiv:2408.01228

The Phantom Menace: Unmasking Privacy Leakages in Vision-Language Models

Vision-Language Models (VLMs) combine visual and textual understanding, rendering them well-suited for diverse tasks like generating image captions and answering visual questions across various domains. However, these capabilities are built upon training on large amount of uncurated data crawled from the web. The latter may include sensitive information that VLMs could memorize and leak, raising significant privacy concerns. In this paper, we assess whether these vulnerabilities exist, focusing on identity leakage. Our study leads to three key findings: (i) VLMs leak identity information, even when the vision-language alignment and the fine-tuning use anonymized data; (ii) context has little influence on identity leakage; (iii) simple, widely used anonymization techniques, like blurring, are not sufficient to address the problem. These findings underscore the urgent need for robust privacy protection strategies when deploying VLMs. Ethical awareness and responsible development practices are essential to mitigate these risks.

arXiv:2407.21174

AI Safety in Practice: Enhancing Adversarial Robustness in Multimodal Image Captioning

Multimodal machine learning models that combine visual and textual data are increasingly being deployed in critical applications, raising significant safety and security concerns due to their vulnerability to adversarial attacks. This paper presents an effective strategy to enhance the robustness of multimodal image captioning models against such attacks. By leveraging the Fast Gradient Sign Method (FGSM) to generate adversarial examples and incorporating adversarial training techniques, we demonstrate improved model robustness on two benchmark datasets: Flickr8k and COCO. Our findings indicate that selectively training only the text decoder of the multimodal architecture shows performance comparable to full adversarial training while offering increased computational efficiency. This targeted approach suggests a balance between robustness and training costs, facilitating the ethical deployment of multimodal AI systems across various domains.

arXiv:2407.20341공식 코드 · Papers with Code

BRIDGE: Bridging Gaps in Image Captioning Evaluation with Stronger Visual Cues

Effectively aligning with human judgment when evaluating machine-generated image captions represents a complex yet intriguing challenge. Existing evaluation metrics like CIDEr or CLIP-Score fall short in this regard as they do not take into account the corresponding image or lack the capability of encoding fine-grained details and penalizing hallucinations. To overcome these issues, in this paper, we propose BRIDGE, a new learnable and reference-free image captioning metric that employs a novel module to map visual features into dense vectors and integrates them into multi-modal pseudo-captions which are built during the evaluation process. This approach results in a multimodal metric that properly incorporates information from the input image without relying on reference captions, bridging the gap between human judgment and machine-generated image captions. Experiments spanning several datasets demonstrate that our proposal achieves state-of-the-art results compared to existing reference-free evaluation scores. Our source code and trained models are publicly available at: https://github.com/aimagelab/bridge-score.

arXiv:2407.18995공식 코드 · Papers with Code

SWIFT: Semantic Watermarking for Image Forgery Thwarting

This paper proposes a novel approach towards image authentication and tampering detection by using watermarking as a communication channel for semantic information. We modify the HiDDeN deep-learning watermarking architecture to embed and extract high-dimensional real vectors representing image captions. Our method improves significantly robustness on both malign and benign edits. We also introduce a local confidence metric correlated with Message Recovery Rate, enhancing the method's practical applicability. This approach bridges the gap between traditional watermarking and passive forensic methods, offering a robust solution for image integrity verification.

arXiv:2407.19795

VolDoGer: LLM-assisted Datasets for Domain Generalization in Vision-Language Tasks

Domain generalizability is a crucial aspect of a deep learning model since it determines the capability of the model to perform well on data from unseen domains. However, research on the domain generalizability of deep learning models for vision-language tasks remains limited, primarily because of the lack of required datasets. To address these challenges, we propose VolDoGer: Vision-Language Dataset for Domain Generalization, a dedicated dataset designed for domain generalization that addresses three vision-language tasks: image captioning, visual question answering, and visual entailment. We constructed VolDoGer by extending LLM-based data annotation techniques to vision-language tasks, thereby alleviating the burden of recruiting human annotators. We evaluated the domain generalizability of various models, ranging from fine-tuned models to a recent multimodal large language model, through VolDoGer.

arXiv:2407.18589

HICEScore: A Hierarchical Metric for Image Captioning Evaluation

Image captioning evaluation metrics can be divided into two categories, reference-based metrics and reference-free metrics. However, reference-based approaches may struggle to evaluate descriptive captions with abundant visual details produced by advanced multimodal large language models, due to their heavy reliance on limited human-annotated references. In contrast, previous reference-free metrics have been proven effective via CLIP cross-modality similarity. Nonetheless, CLIP-based metrics, constrained by their solution of global image-text compatibility, often have a deficiency in detecting local textual hallucinations and are insensitive to small visual objects. Besides, their single-scale designs are unable to provide an interpretable evaluation process such as pinpointing the position of caption mistakes and identifying visual regions that have not been described. To move forward, we propose a novel reference-free metric for image captioning evaluation, dubbed Hierarchical Image Captioning Evaluation Score (HICE-S). By detecting local visual regions and textual phrases, HICE-S builds an interpretable hierarchical scoring mechanism, breaking through the barriers of the single-scale structure of existing reference-free metrics. Comprehensive experiments indicate that our proposed metric achieves the SOTA performance on several benchmarks, outperforming existing reference-free metrics like CLIP-S and PAC-S, and reference-based metrics like METEOR and CIDEr. Moreover, several case studies reveal that the assessment process of HICE-S on detailed captions closely resembles interpretable human judgments.Our code is available at https://github.com/joeyz0z/HICE.

arXiv:2407.16526

Imperfect Vision Encoders: Efficient and Robust Tuning for Vision-Language Models

Vision language models (VLMs) demonstrate impressive capabilities in visual question answering and image captioning, acting as a crucial link between visual and language models. However, existing open-source VLMs heavily rely on pretrained and frozen vision encoders (such as CLIP). Despite CLIP's robustness across diverse domains, it still exhibits non-negligible image understanding errors. These errors propagate to the VLM responses, resulting in sub-optimal performance. In our work, we propose an efficient and robust method for updating vision encoders within VLMs. Our approach selectively and locally updates encoders, leading to substantial performance improvements on data where previous mistakes occurred, while maintaining overall robustness. Furthermore, we demonstrate the effectiveness of our method during continual few-shot updates. Theoretical grounding, generality, and computational efficiency characterize our approach.

arXiv:2407.14720

Downstream-Pretext Domain Knowledge Traceback for Active Learning

Active learning (AL) is designed to construct a high-quality labeled dataset by iteratively selecting the most informative samples. Such sampling heavily relies on data representation, while recently pre-training is popular for robust feature learning. However, as pre-training utilizes low-level pretext tasks that lack annotation, directly using pre-trained representation in AL is inadequate for determining the sampling score. To address this problem, we propose a downstream-pretext domain knowledge traceback (DOKT) method that traces the data interactions of downstream knowledge and pre-training guidance for selecting diverse and instructive samples near the decision boundary. DOKT consists of a traceback diversity indicator and a domain-based uncertainty estimator. The diversity indicator constructs two feature spaces based on the pre-training pretext model and the downstream knowledge from annotation, by which it locates the neighbors of unlabeled data from the downstream space in the pretext space to explore the interaction of samples. With this mechanism, DOKT unifies the data relations of low-level and high-level representations to estimate traceback diversity. Next, in the uncertainty estimator, domain mixing is designed to enforce perceptual perturbing to unlabeled samples with similar visual patches in the pretext space. Then the divergence of perturbed samples is measured to estimate the domain uncertainty. As a result, DOKT selects the most diverse and important samples based on these two modules. The experiments conducted on ten datasets show that our model outperforms other state-of-the-art methods and generalizes well to various application scenarios such as semantic segmentation and image captioning.

arXiv:2407.15488공식 코드 · Papers with Code

DiffX: Guide Your Layout to Cross-Modal Generative Modeling

Diffusion models have made significant strides in language-driven and layout-driven image generation. However, most diffusion models are limited to visible RGB image generation. In fact, human perception of the world is enriched by diverse viewpoints, such as chromatic contrast, thermal illumination, and depth information. In this paper, we introduce a novel diffusion model for general layout-guided cross-modal generation, called DiffX. Notably, our DiffX presents a compact and effective cross-modal generative modeling pipeline, which conducts diffusion and denoising processes in the modality-shared latent space. Moreover, we introduce the Joint-Modality Embedder (JME) to enhance the interaction between layout and text conditions by incorporating a gated attention mechanism. To facilitate the user-instructed training, we construct the cross-modal image datasets with detailed text captions by the Large-Multimodal Model (LMM) and our human-in-the-loop refinement. Through extensive experiments, our DiffX demonstrates robustness in cross-modal ''RGB+X'' image generation on FLIR, MFNet, and COME15K datasets, guided by various layout conditions. Meanwhile, it shows the strong potential for the adaptive generation of ``RGB+X+Y(+Z)'' images or more diverse modalities on FLIR, MFNet, COME15K, and MCXFace datasets. To our knowledge, DiffX is the first model for layout-guided cross-modal image generation. Our code and constructed cross-modal image datasets are available at https://github.com/zeyuwang-zju/DiffX.

arXiv:2407.15295

VideoGameBunny: Towards vision assistants for video games

Large multimodal models (LMMs) hold substantial promise across various domains, from personal assistance in daily tasks to sophisticated applications like medical diagnostics. However, their capabilities have limitations in the video game domain, such as challenges with scene understanding, hallucinations, and inaccurate descriptions of video game content, especially in open-source models. This paper describes the development of VideoGameBunny, a LLaVA-style model based on Bunny, specifically tailored for understanding images from video games. We release intermediate checkpoints, training logs, and an extensive dataset comprising 185,259 video game images from 413 titles, along with 389,565 image-instruction pairs that include image captions, question-answer pairs, and a JSON representation of 16 elements of 136,974 images. Our experiments show that our high quality game-related data has the potential to make a relatively small model outperform the much larger state-of-the-art model LLaVa-1.6-34b (which has more than 4x the number of parameters). Our study paves the way for future research in video game understanding on tasks such as playing, commentary, and debugging. Code and data are available at https://videogamebunny.github.io/

arXiv:2407.14177

EVLM: An Efficient Vision-Language Model for Visual Understanding

In the field of multi-modal language models, the majority of methods are built on an architecture similar to LLaVA. These models use a single-layer ViT feature as a visual prompt, directly feeding it into the language models alongside textual tokens. However, when dealing with long sequences of visual signals or inputs such as videos, the self-attention mechanism of language models can lead to significant computational overhead. Additionally, using single-layer ViT features makes it challenging for large language models to perceive visual signals fully. This paper proposes an efficient multi-modal language model to minimize computational costs while enabling the model to perceive visual signals as comprehensively as possible. Our method primarily includes: (1) employing cross-attention to image-text interaction similar to Flamingo. (2) utilize hierarchical ViT features. (3) introduce the Mixture of Experts (MoE) mechanism to enhance model effectiveness. Our model achieves competitive scores on public multi-modal benchmarks and performs well in tasks such as image captioning and video captioning.

arXiv:2407.12753

LookupViT: Compressing visual information to a limited number of tokens

Vision Transformers (ViT) have emerged as the de-facto choice for numerous industry grade vision solutions. But their inference cost can be prohibitive for many settings, as they compute self-attention in each layer which suffers from quadratic computational complexity in the number of tokens. On the other hand, spatial information in images and spatio-temporal information in videos is usually sparse and redundant. In this work, we introduce LookupViT, that aims to exploit this information sparsity to reduce ViT inference cost. LookupViT provides a novel general purpose vision transformer block that operates by compressing information from higher resolution tokens to a fixed number of tokens. These few compressed tokens undergo meticulous processing, while the higher-resolution tokens are passed through computationally cheaper layers. Information sharing between these two token sets is enabled through a bidirectional cross-attention mechanism. The approach offers multiple advantages - (a) easy to implement on standard ML accelerators (GPUs/TPUs) via standard high-level operators, (b) applicable to standard ViT and its variants, thus generalizes to various tasks, (c) can handle different tokenization and attention approaches. LookupViT also offers flexibility for the compressed tokens, enabling performance-computation trade-offs in a single trained model. We show LookupViT's effectiveness on multiple domains - (a) for image-classification (ImageNet-1K and ImageNet-21K), (b) video classification (Kinetics400 and Something-Something V2), (c) image captioning (COCO-Captions) with a frozen encoder. LookupViT provides $2\times$ reduction in FLOPs while upholding or improving accuracy across these domains. In addition, LookupViT also demonstrates out-of-the-box robustness and generalization on image classification (ImageNet-C,R,A,O), improving by up to $4\%$ over ViT.

arXiv:2407.11449공식 코드 · Papers with Code

Controllable Contextualized Image Captioning: Directing the Visual Narrative through User-Defined Highlights

Contextualized Image Captioning (CIC) evolves traditional image captioning into a more complex domain, necessitating the ability for multimodal reasoning. It aims to generate image captions given specific contextual information. This paper further introduces a novel domain of Controllable Contextualized Image Captioning (Ctrl-CIC). Unlike CIC, which solely relies on broad context, Ctrl-CIC accentuates a user-defined highlight, compelling the model to tailor captions that resonate with the highlighted aspects of the context. We present two approaches, Prompting-based Controller (P-Ctrl) and Recalibration-based Controller (R-Ctrl), to generate focused captions. P-Ctrl conditions the model generation on highlight by prepending captions with highlight-driven prefixes, whereas R-Ctrl tunes the model to selectively recalibrate the encoder embeddings for highlighted tokens. Additionally, we design a GPT-4V empowered evaluator to assess the quality of the controlled captions alongside standard assessment methods. Extensive experimental results demonstrate the efficient and effective controllability of our method, charting a new direction in achieving user-adaptive image captioning. Code is available at https://github.com/ShunqiM/Ctrl-CIC .

arXiv:2407.11393공식 코드 · Papers with Code

CIC-BART-SSA: Controllable Image Captioning with Structured Semantic Augmentation

Controllable Image Captioning (CIC) aims at generating natural language descriptions for an image, conditioned on information provided by end users, e.g., regions, entities or events of interest. However, available image-language datasets mainly contain captions that describe the entirety of an image, making them ineffective for training CIC models that can potentially attend to any subset of regions or relationships. To tackle this challenge, we propose a novel, fully automatic method to sample additional focused and visually grounded captions using a unified structured semantic representation built on top of the existing set of captions associated with an image. We leverage Abstract Meaning Representation (AMR), a cross-lingual graph-based semantic formalism, to encode all possible spatio-semantic relations between entities, beyond the typical spatial-relations-only focus of current methods. We use this Structured Semantic Augmentation (SSA) framework to augment existing image-caption datasets with the grounded controlled captions, increasing their spatial and semantic diversity and focal coverage. We then develop a new model, CIC-BART-SSA, specifically tailored for the CIC task, that sources its control signals from SSA-diversified datasets. We empirically show that, compared to SOTA CIC models, CIC-BART-SSA generates captions that are superior in diversity and text quality, are competitive in controllability, and, importantly, minimize the gap between broad and highly focused controlled captioning performance by efficiently generalizing to the challenging highly focused scenarios. Code is available at https://github.com/SamsungLabs/CIC-BART-SSA.

arXiv:2407.09556

Explainable Image Captioning using CNN- CNN architecture and Hierarchical Attention

Image captioning is a technology that produces text-based descriptions for an image. Deep learning-based solutions built on top of feature recognition may very well serve the purpose. But as with any other machine learning solution, the user understanding in the process of caption generation is poor and the model does not provide any explanation for its predictions and hence the conventional methods are also referred to as Black-Box methods. Thus, an approach where the model's predictions are trusted by the user is needed to appreciate interoperability. Explainable AI is an approach where a conventional method is approached in a way that the model or the algorithm's predictions can be explainable and justifiable. Thus, this article tries to approach image captioning using Explainable AI such that the resulting captions generated by the model can be Explained and visualized. A newer architecture with a CNN decoder and hierarchical attention concept has been used to increase speed and accuracy of caption generation. Also, incorporating explainability to a model makes it more trustable when used in an application. The model is trained and evaluated using MSCOCO dataset and both quantitative and qualitative results are presented in this article.

arXiv:2407.09548

Towards Temporal Change Explanations from Bi-Temporal Satellite Images

Explaining temporal changes between satellite images taken at different times is important for urban planning and environmental monitoring. However, manual dataset construction for the task is costly, so human-AI collaboration is promissing. Toward the direction, in this paper, we investigate the ability of Large-scale Vision-Language Models (LVLMs) to explain temporal changes between satellite images. While LVLMs are known to generate good image captions, they receive only a single image as input. To deal with a par of satellite images as input, we propose three prompting methods. Through human evaluation, we found the effectiveness of our step-by-step reasoning based prompting.

arXiv:2407.09495

Image captioning in different languages

This short position paper provides a manually curated list of non-English image captioning datasets (as of May 2024). Through this list, we can observe the dearth of datasets in different languages: only 23 different languages are represented. With the addition of the Crossmodal-3600 dataset (Thapliyal et al., 2022, 36 languages) this number increases somewhat, but still this number is small compared to the +/-500 institutional languages that are out there. This paper closes with some open questions for the field of Vision & Language.

논문 ID: 1783.36122

CropCap: Embedding Visual Cross-Partition Dependency for Image Captioning

Transformer-based approaches to image captioning have shown great success by utilizing long-term dependency for visual embedding. However, their coarse long-term dependency, using the multi-head self-attention mechanism to capture the contextual interactions between the visual tokens on the time step and (or) embedded dimension, fail to distinguish fine-grained features of local partition. In this case, some similar features are captured, which leads to feature redundancy that decreases the performance. To respond to this issue, this paper proposes a novel image captioner embedding visual cross-partition dependency, dubbed CropCap. Specifically, the visual sequence generated from the Swin Transformer-based pre-embedding network is fed into the proposed cross-partition dependency module to refinedly model the interaction between partial representations on both the time step and embedded dimension. Furthermore, we formulaically reason the proposed cross-partition dependency, and theoretically prove its correctness. Extensive comparisons on the benchmark MS-COCO dataset demonstrated the effectiveness addressing the information redundancy issue, and verified the superior performance of our method.