MMCOMPOSITION: Revisiting the Compositionality of Pre-trained Vision-Language Models

The advent of large Vision-Language Models (VLMs) has significantly advanced multimodal understanding, enabling more sophisticated and accurate integration of visual and textual information across various tasks, including image and video captioning, visual question answering, and cross-modal retrieval. Despite VLMs' superior capabilities, researchers lack a comprehensive understanding of their compositionality -- the ability to understand and produce novel combinations of known visual and textual components. Prior benchmarks provide only a relatively rough compositionality evaluation from the perspectives of objects, relations, and attributes while neglecting deeper reasoning about object interactions, counting, and complex compositions. However, compositionality is a critical ability that facilitates coherent reasoning and understanding across modalities for VLMs. To address this limitation, we propose MMCOMPOSITION, a novel human-annotated benchmark for comprehensively and accurately evaluating VLMs' compositionality. Our proposed benchmark serves as a complement to these earlier works. With MMCOMPOSITION, we can quantify and explore the compositionality of the mainstream VLMs. Surprisingly, we find GPT-4o's compositionality inferior to the best open-source model, and we analyze the underlying reasons. Our experimental analysis reveals the limitations of VLMs in fine-grained compositional perception and reasoning, and points to areas for improvement in VLM design and training. Resources available at: https://hanghuacs.github.io/MMComposition/

PromptFix: You Prompt and We Fix the Photo

Diffusion models equipped with language models demonstrate excellent controllability in image generation tasks, allowing image processing to adhere to human instructions. However, the lack of diverse instruction-following data hampers the development of models that effectively recognize and execute user-customized instructions, particularly in low-level tasks. Moreover, the stochastic nature of the diffusion process leads to deficiencies in image generation or editing tasks that require the detailed preservation of the generated images. To address these limitations, we propose PromptFix, a comprehensive framework that enables diffusion models to follow human instructions to perform a wide variety of image-processing tasks. First, we construct a large-scale instruction-following dataset that covers comprehensive image-processing tasks, including low-level tasks, image editing, and object creation. Next, we propose a high-frequency guidance sampling method to explicitly control the denoising process and preserve high-frequency details in unprocessed areas. Finally, we design an auxiliary prompting adapter, utilizing Vision-Language Models (VLMs) to enhance text prompts and improve the model's task generalization. Experimental results show that PromptFix outperforms previous methods in various image-processing tasks. Our proposed model also achieves comparable inference efficiency with these baseline models and exhibits superior zero-shot capabilities in blind restoration and combination tasks. The dataset and code will be aviliable at https://github.com/yeates/PromptFix.

FINEMATCH: Aspect-based Fine-grained Image and Text Mismatch Detection and Correction

Recent progress in large-scale pre-training has led to the development of advanced vision-language models (VLMs) with remarkable proficiency in comprehending and generating multimodal content. Despite the impressive ability to perform complex reasoning for VLMs, current models often struggle to effectively and precisely capture the compositional information on both the image and text sides. To address this, we propose FineMatch, a new aspect-based fine-grained text and image matching benchmark, focusing on text and image mismatch detection and correction. This benchmark introduces a novel task for boosting and evaluating the VLMs' compositionality for aspect-based fine-grained text and image matching. In this task, models are required to identify mismatched aspect phrases within a caption, determine the aspect's class, and propose corrections for an image-text pair that may contain between 0 and 3 mismatches. To evaluate the models' performance on this new task, we propose a new evaluation metric named ITM-IoU for which our experiments show a high correlation to human evaluation. In addition, we also provide a comprehensive experimental analysis of existing mainstream VLMs, including fully supervised learning and in-context learning settings. We have found that models trained on FineMatch demonstrate enhanced proficiency in detecting fine-grained text and image mismatches. Moreover, models (e.g., GPT-4V, Gemini Pro Vision) with strong abilities to perform multimodal in-context learning are not as skilled at fine-grained compositional image and text matching analysis. With FineMatch, we are able to build a system for text-to-image generation hallucination detection and correction.

BattleAgent: Multi-modal Dynamic Emulation on Historical Battles to Complement Historical Analysis

This paper presents BattleAgent, an emulation system that combines the Large Vision-Language Model and Multi-agent System. This novel system aims to simulate complex dynamic interactions among multiple agents, as well as between agents and their environments, over a period of time. It emulates both the decision-making processes of leaders and the viewpoints of ordinary participants, such as soldiers. The emulation showcases the current capabilities of agents, featuring fine-grained multi-modal interactions between agents and landscapes. It develops customizable agent structures to meet specific situational requirements, for example, a variety of battle-related activities like scouting and trench digging. These components collaborate to recreate historical events in a lively and comprehensive manner while offering insights into the thoughts and feelings of individuals from diverse viewpoints. The technological foundations of BattleAgent establish detailed and immersive settings for historical battles, enabling individual agents to partake in, observe, and dynamically respond to evolving battle scenarios. This methodology holds the potential to substantially deepen our understanding of historical events, particularly through individual accounts. Such initiatives can also aid historical research, as conventional historical narratives often lack documentation and prioritize the perspectives of decision-makers, thereby overlooking the experiences of ordinary individuals. BattelAgent illustrates AI's potential to revitalize the human aspect in crucial social events, thereby fostering a more nuanced collective understanding and driving the progressive development of human society.

V2Xum-LLM: Cross-Modal Video Summarization with Temporal Prompt Instruction Tuning

Video summarization aims to create short, accurate, and cohesive summaries of longer videos. Despite the existence of various video summarization datasets, a notable limitation is their limited amount of source videos, which hampers the effective fine-tuning of advanced large vision-language models (VLMs). Additionally, most existing datasets are created for video-to-video summarization, overlooking the contemporary need for multimodal video content summarization. Recent efforts have been made to expand from unimodal to multimodal video summarization, categorizing the task into three sub-tasks based on the summary's modality: video-to-video (V2V), video-to-text (V2T), and a combination of video and text summarization (V2VT). However, the textual summaries in previous multimodal datasets are inadequate. To address these issues, we introduce Instruct-V2Xum, a cross-modal video summarization dataset featuring 30,000 diverse videos sourced from YouTube, with lengths ranging from 40 to 940 seconds and an average summarization ratio of 16.39\%. Each video summary in Instruct-V2Xum is paired with a textual summary that references specific frame indexes, facilitating the generation of aligned video and textual summaries. In addition, we propose a new video summarization framework named V2Xum-LLM. V2Xum-LLM, specifically V2Xum-LLaMA in this study, is the first framework that unifies different video summarization tasks into one large language model's (LLM) text decoder and achieves task-controllable video summarization with temporal prompts and task instructions. Experiments show that V2Xum-LLaMA outperforms strong baseline models on multiple video summarization tasks. Furthermore, we propose an enhanced evaluation metric for V2V and V2VT summarization tasks.

Emo-Avatar: Efficient Monocular Video Style Avatar through Texture Rendering

Artistic video portrait generation is a significant and sought-after task in the fields of computer graphics and vision. While various methods have been developed that integrate NeRFs or StyleGANs with instructional editing models for creating and editing drivable portraits, these approaches face several challenges. They often rely heavily on large datasets, require extensive customization processes, and frequently result in reduced image quality. To address the above problems, we propose the Efficient Monotonic Video Style Avatar (Emo-Avatar) through deferred neural rendering that enhances StyleGAN's capacity for producing dynamic, drivable portrait videos. We proposed a two-stage deferred neural rendering pipeline. In the first stage, we utilize few-shot PTI initialization to initialize the StyleGAN generator through several extreme poses sampled from the video to capture the consistent representation of aligned faces from the target portrait. In the second stage, we propose a Laplacian pyramid for high-frequency texture sampling from UV maps deformed by dynamic flow of expression for motion-aware texture prior integration to provide torso features to enhance StyleGAN's ability to generate complete and upper body for portrait video rendering. Emo-Avatar reduces style customization time from hours to merely 5 minutes compared with existing methods. In addition, Emo-Avatar requires only a single reference image for editing and employs region-aware contrastive learning with semantic invariant CLIP guidance, ensuring consistent high-resolution output and identity preservation. Through both quantitative and qualitative assessments, Emo-Avatar demonstrates superior performance over existing methods in terms of training efficiency, rendering quality and editability in self- and cross-reenactment.

VideoXum: Cross-modal Visual and Textural Summarization of Videos

Video summarization aims to distill the most important information from a source video to produce either an abridged clip or a textual narrative. Traditionally, different methods have been proposed depending on whether the output is a video or text, thus ignoring the correlation between the two semantically related tasks of visual summarization and textual summarization. We propose a new joint video and text summarization task. The goal is to generate both a shortened video clip along with the corresponding textual summary from a long video, collectively referred to as a cross-modal summary. The generated shortened video clip and text narratives should be semantically well aligned. To this end, we first build a large-scale human-annotated dataset -- VideoXum (X refers to different modalities). The dataset is reannotated based on ActivityNet. After we filter out the videos that do not meet the length requirements, 14,001 long videos remain in our new dataset. Each video in our reannotated dataset has human-annotated video summaries and the corresponding narrative summaries. We then design a novel end-to-end model -- VTSUM-BILP to address the challenges of our proposed task. Moreover, we propose a new metric called VT-CLIPScore to help evaluate the semantic consistency of cross-modality summary. The proposed model achieves promising performance on this new task and establishes a benchmark for future research.

PromptCap: Prompt-Guided Task-Aware Image Captioning

Image captioning aims to describe an image with a natural language sentence, allowing powerful language models to understand images. The framework of combining image captioning with language models has been successful on various vision-language tasks. However, an image contains much more information than a single sentence, leading to underspecification of which visual entities should be described in the caption sentence. For example, when performing visual questioning answering (VQA), generic image captions often miss visual details that are essential for the language model to answer correctly. To address this challenge, we propose PromptCap, a captioning model that takes a natural-language prompt to control the contents of the generated caption. The prompt contains a question that the caption should help to answer, and also supports taking auxiliary text inputs such as scene text within the image itself. To finetune a general image caption model for prompt-guided captioning, we propose a pipeline to synthesize and filter training examples with GPT-3 and existing VQA datasets. For evaluation, we start with an existing pipeline in which a language model is prompted with image captions to carry out VQA. With the same language model, a higher QA accuracy shows that our generated captions are more relevant to the question prompts. PromptCap outperforms generic captions by a large margin on a variety of VQA tasks and achieves the state-of-the-art accuracy of 58.8 % on OK-VQA and 58.0 % on A-OKVQA. Zero-shot experiments on WebQA show that PromptCap generalizes well to unseen domains.

Fine-tuning Pre-trained Language Models with Noise Stability Regularization

The advent of large-scale pre-trained language models has contributed greatly to the recent progress in natural language processing. Many state-of-the-art language models are first trained on a large text corpus and then fine-tuned on downstream tasks. Despite its recent success and wide adoption, fine-tuning a pre-trained language model often suffers from overfitting, which leads to poor generalizability due to the extremely high complexity of the model and the limited training samples from downstream tasks. To address this problem, we propose a novel and effective fine-tuning framework, named Layerwise Noise Stability Regularization (LNSR). Specifically, we propose to inject the standard Gaussian noise or In-manifold noise and regularize hidden representations of the fine-tuned model. We first provide theoretical analyses to support the efficacy of our method. We then demonstrate the advantages of the proposed method over other state-of-the-art algorithms including L2-SP, Mixout and SMART. While these previous works only verify the effectiveness of their methods on relatively simple text classification tasks, we also verify the effectiveness of our method on question answering tasks, where the target problem is much more difficult and more training examples are available. Furthermore, extensive experimental results indicate that the proposed algorithm can not only enhance the in-domain performance of the language models but also improve the domain generalization performance on out-of-domain data.

Noise Stability Regularization for Improving BERT Fine-tuning

Fine-tuning pre-trained language models such as BERT has become a common practice dominating leaderboards across various NLP tasks. Despite its recent success and wide adoption, this process is unstable when there are only a small number of training samples available. The brittleness of this process is often reflected by the sensitivity to random seeds. In this paper, we propose to tackle this problem based on the noise stability property of deep nets, which is investigated in recent literature (Arora et al., 2018; Sanyal et al., 2020). Specifically, we introduce a novel and effective regularization method to improve fine-tuning on NLP tasks, referred to as Layer-wise Noise Stability Regularization (LNSR). We extend the theories about adding noise to the input and prove that our method gives a stabler regularization effect. We provide supportive evidence by experimentally confirming that well-performing models show a low sensitivity to noise and fine-tuning with LNSR exhibits clearly higher generalizability and stability. Furthermore, our method also demonstrates advantages over other state-of-the-art algorithms including L2-SP (Li et al., 2018), Mixout (Lee et al., 2020) and SMART (Jiang et al., 2020).