Recently, Multimodal Large Language Model (MLLM) represented by GPT-4V has been a new rising research hotspot, which uses powerful Large Language Models (LLMs) as a brain to perform multimodal tasks. The surprising emergent capabilities of MLLM, such as writing stories based on images and OCR-free math reasoning, are rare in traditional multimodal methods, suggesting a potential path to artificial general intelligence. To this end, both academia and industry have endeavored to develop MLLMs that can compete with or even better than GPT-4V, pushing the limit of research at a surprising speed. In this paper, we aim to trace and summarize the recent progress of MLLMs. First of all, we present the basic formulation of MLLM and delineate its related concepts, including architecture, training strategy and data, as well as evaluation. Then, we introduce research topics about how MLLMs can be extended to support more granularity, modalities, languages, and scenarios. We continue with multimodal hallucination and extended techniques, including Multimodal ICL (M-ICL), Multimodal CoT (M-CoT), and LLM-Aided Visual Reasoning (LAVR). To conclude the paper, we discuss existing challenges and point out promising research directions. In light of the fact that the era of MLLM has only just begun, we will keep updating this survey and hope it can inspire more research. An associated GitHub link collecting the latest papers is available at https://github.com/BradyFU/Awesome-Multimodal-Large-Language-Models.
Despite the successes of recent developments in visual AI, different shortcomings still exist; from missing exact logical reasoning, to abstract generalization abilities, to understanding complex and noisy scenes. Unfortunately, existing benchmarks, were not designed to capture more than a few of these aspects. Whereas deep learning datasets focus on visually complex data but simple visual reasoning tasks, inductive logic datasets involve complex logical learning tasks, however, lack the visual component. To address this, we propose the diagnostic visual logical learning dataset, V-LoL, that seamlessly combines visual and logical challenges. Notably, we introduce the first instantiation of V-LoL, V-LoL-Train, - a visual rendition of a classic benchmark in symbolic AI, the Michalski train problem. By incorporating intricate visual scenes and flexible logical reasoning tasks within a versatile framework, V-LoL-Train provides a platform for investigating a wide range of visual logical learning challenges. We evaluate a variety of AI systems including traditional symbolic AI, neural AI, as well as neuro-symbolic AI. Our evaluations demonstrate that even SOTA AI faces difficulties in dealing with visual logical learning challenges, highlighting unique advantages and limitations of each methodology. Overall, V-LoL opens up new avenues for understanding and enhancing current abilities in visual logical learning for AI systems.
Model-based reasoning agents are ill-equipped to act in novel situations in which their model of the environment no longer sufficiently represents the world. We propose HYDRA - a framework for designing model-based agents operating in mixed discrete-continuous worlds, that can autonomously detect when the environment has evolved from its canonical setup, understand how it has evolved, and adapt the agents' models to perform effectively. HYDRA is based upon PDDL+, a rich modeling language for planning in mixed, discrete-continuous environments. It augments the planning module with visual reasoning, task selection, and action execution modules for closed-loop interaction with complex environments. HYDRA implements a novel meta-reasoning process that enables the agent to monitor its own behavior from a variety of aspects. The process employs a diverse set of computational methods to maintain expectations about the agent's own behavior in an environment. Divergences from those expectations are useful in detecting when the environment has evolved and identifying opportunities to adapt the underlying models. HYDRA builds upon ideas from diagnosis and repair and uses a heuristics-guided search over model changes such that they become competent in novel conditions. The HYDRA framework has been used to implement novelty-aware agents for three diverse domains - CartPole++ (a higher dimension variant of a classic control problem), Science Birds (an IJCAI competition problem), and PogoStick (a specific problem domain in Minecraft). We report empirical observations from these domains to demonstrate the efficacy of various components in the novelty meta-reasoning process.
Large language models (LLMs) have made significant advancements in natural language understanding. However, through that enormous semantic representation that the LLM has learnt, is it somehow possible for it to understand images as well? This work investigates this question. To enable the LLM to process images, we convert them into a representation given by Scalable Vector Graphics (SVG). To study what the LLM can do with this XML-based textual description of images, we test the LLM on three broad computer vision tasks: (i) visual reasoning and question answering, (ii) image classification under distribution shift, few-shot learning, and (iii) generating new images using visual prompting. Even though we do not naturally associate LLMs with any visual understanding capabilities, our results indicate that the LLM can often do a decent job in many of these tasks, potentially opening new avenues for research into LLMs' ability to understand image data. Our code, data, and models can be found here https://github.com/mu-cai/svg-llm.
Vision-language models (VLMs) are impactful in part because they can be applied to a variety of visual understanding tasks in a zero-shot fashion, without any fine-tuning. We study $\textit{generative VLMs}$ that are trained for next-word generation given an image. We explore their zero-shot performance on the illustrative task of image-text retrieval across 8 popular vision-language benchmarks. Our first observation is that they can be repurposed for discriminative tasks (such as image-text retrieval) by simply computing the match score of generating a particular text string given an image. We call this probabilistic score the $\textit{Visual Generative Pre-Training Score}$ (VisualGPTScore). While the VisualGPTScore produces near-perfect accuracy on some retrieval benchmarks, it yields poor accuracy on others. We analyze this behavior through a probabilistic lens, pointing out that some benchmarks inadvertently capture unnatural language distributions by creating adversarial but unlikely text captions. In fact, we demonstrate that even a "blind" language model that ignores any image evidence can sometimes outperform all prior art, reminiscent of similar challenges faced by the visual-question answering (VQA) community many years ago. We derive a probabilistic post-processing scheme that controls for the amount of linguistic bias in generative VLMs at test time without having to retrain or fine-tune the model. We show that the VisualGPTScore, when appropriately debiased, is a strong zero-shot baseline for vision-language understanding, oftentimes producing state-of-the-art accuracy.
Recent vision-language models have achieved tremendous advances. However, their computational costs are also escalating dramatically, making model acceleration exceedingly critical. To pursue more efficient vision-language Transformers, this paper introduces Cross-Guided Ensemble of Tokens (CrossGET), a general acceleration framework for vision-language Transformers. This framework adaptively combines tokens in real-time during inference, significantly reducing computational costs while maintaining high performance. CrossGET features two primary innovations: 1) Cross-Guided Matching and Ensemble. CrossGET leverages cross-modal guided token matching and ensemble to effectively utilize cross-modal information, achieving wider applicability across both modality-independent models, e.g., CLIP, and modality-dependent ones, e.g., BLIP2. 2) Complete-Graph Soft Matching. CrossGET introduces an algorithm for the token-matching mechanism, ensuring reliable matching results while facilitating parallelizability and high efficiency. Extensive experiments have been conducted on various vision-language tasks, such as image-text retrieval, visual reasoning, image captioning, and visual question answering. The performance on both classic multimodal architectures and emerging multimodal LLMs demonstrates the framework's effectiveness and versatility. The code is available at https://github.com/sdc17/CrossGET.
While pretraining on large-scale image-text data from the Web has facilitated rapid progress on many vision-and-language (V&L) tasks, recent work has demonstrated that pretrained models lack "fine-grained" understanding, such as the ability to recognise relationships, verbs, and numbers in images. This has resulted in an increased interest in the community to either develop new benchmarks or models for such capabilities. To better understand and quantify progress in this direction, we investigate four competitive V&L models on four fine-grained benchmarks. Through our analysis, we find that X-VLM (Zeng et al., 2022) consistently outperforms other baselines, and that modelling innovations can impact performance more than scaling Web data, which even degrades performance sometimes. Through a deeper investigation of X-VLM, we highlight the importance of both novel losses and rich data sources for learning fine-grained skills. Finally, we inspect training dynamics, and discover that for some tasks, performance peaks early in training or significantly fluctuates, never converging.
The ability to judge whether a caption correctly describes an image is a critical part of vision-language understanding. However, state-of-the-art models often misinterpret the correctness of fine-grained details, leading to errors in outputs such as hallucinating objects in generated captions or poor compositional reasoning. In this work, we explore Token-Level Confidence, or TLC, as a simple yet surprisingly effective method to assess caption correctness. Specifically, we fine-tune a vision-language model on image captioning, input an image and proposed caption to the model, and aggregate either algebraic or learned token confidences over words or sequences to estimate image-caption consistency. Compared to sequence-level scores from pretrained models, TLC with algebraic confidence measures achieves a relative improvement in accuracy by 10% on verb understanding in SVO-Probes and outperforms prior state-of-the-art in image and group scores for compositional reasoning in Winoground by a relative 37% and 9%, respectively. When training data are available, a learned confidence estimator provides further improved performance, reducing object hallucination rates in MS COCO Captions by a relative 30% over the original model and setting a new state-of-the-art.
Vision and language models (VLMs) have demonstrated remarkable zero-shot (ZS) performance in a variety of tasks. However, recent works have shown that even the best VLMs struggle to capture aspects of compositional scene understanding, such as object attributes, relations, and action states. In contrast, obtaining structured annotations, such as scene graphs (SGs), that could improve these models is time-consuming and costly, and thus cannot be used on a large scale. Here we ask whether small SG datasets can provide sufficient information for enhancing structured understanding of pretrained VLMs. We show that it is indeed possible to improve VLMs when learning from SGs by integrating components that incorporate structured information into both visual and textual representations. For the visual side, we incorporate a special "SG Component" in the image transformer trained to predict SG information, while for the textual side, we utilize SGs to generate fine-grained captions that highlight different compositional aspects of the scene. Our method improves the performance of several popular VLMs on multiple VL datasets with only a mild degradation in ZS capabilities.
Large language models (LLMs) have demonstrated significant universal capabilities as few/zero-shot learners in various tasks due to their pre-training on vast amounts of text data, as exemplified by GPT-3, which boosted to InstrctGPT and ChatGPT, effectively following natural language instructions to accomplish real-world tasks. In this paper, we propose to introduce instruction tuning into multi-modal models, motivated by the Flamingo model's upstream interleaved format pretraining dataset. We adopt a similar approach to construct our MultI-Modal In-Context Instruction Tuning (MIMIC-IT) dataset. We then introduce Otter, a multi-modal model based on OpenFlamingo (open-sourced version of DeepMind's Flamingo), trained on MIMIC-IT and showcasing improved instruction-following ability and in-context learning. We also optimize OpenFlamingo's implementation for researchers, democratizing the required training resources from 1$\times$ A100 GPU to 4$\times$ RTX-3090 GPUs, and integrate both OpenFlamingo and Otter into Huggingface Transformers for more researchers to incorporate the models into their customized training and inference pipelines.
Humans can naturally reason from superficial state differences (e.g. ground wetness) to transformations descriptions (e.g. raining) according to their life experience. In this paper, we propose a new visual reasoning task to test this transformation reasoning ability in real-world scenarios, called \textbf{V}isual \textbf{T}ransformation \textbf{T}elling (VTT). Given a series of states (i.e. images), VTT requires to describe the transformation occurring between every two adjacent states. Different from existing visual reasoning tasks that focus on surface state reasoning, the advantage of VTT is that it captures the underlying causes, e.g. actions or events, behind the differences among states. We collect a novel dataset to support the study of transformation reasoning from two existing instructional video datasets, CrossTask and COIN, comprising 13,547 samples. Each sample involves the key state images along with their transformation descriptions. Our dataset covers diverse real-world activities, providing a rich resource for training and evaluation. To construct an initial benchmark for VTT, we test several models, including traditional visual storytelling methods (CST, GLACNet, Densecap) and advanced multimodal large language models (LLaVA v1.5-7B, Qwen-VL-chat, Gemini Pro Vision, GPT-4o, and GPT-4). Experimental results reveal that even state-of-the-art models still face challenges in VTT, highlighting substantial areas for improvement.
Most existing visual reasoning tasks, such as CLEVR in VQA, ignore an important factor, i.e.~transformation. They are solely defined to test how well machines understand concepts and relations within static settings, like one image. Such \textbf{state driven} visual reasoning has limitations in reflecting the ability to infer the dynamics between different states, which has shown to be equally important for human cognition in Piaget's theory. To tackle this problem, we propose a novel \textbf{transformation driven} visual reasoning (TVR) task. Given both the initial and final states, the target becomes to infer the corresponding intermediate transformation. Following this definition, a new synthetic dataset namely TRANCE is first constructed on the basis of CLEVR, including three levels of settings, i.e.~Basic (single-step transformation), Event (multi-step transformation), and View (multi-step transformation with variant views). Next, we build another real dataset called TRANCO based on COIN, to cover the loss of transformation diversity on TRANCE. Inspired by human reasoning, we propose a three-staged reasoning framework called TranNet, including observing, analyzing, and concluding, to test how recent advanced techniques perform on TVR. Experimental results show that the state-of-the-art visual reasoning models perform well on Basic, but are still far from human-level intelligence on Event, View, and TRANCO. We believe the proposed new paradigm will boost the development of machine visual reasoning. More advanced methods and new problems need to be investigated in this direction. The resource of TVR is available at \url{https://hongxin2019.github.io/TVR/}.
The recent GPT-4 has demonstrated extraordinary multi-modal abilities, such as directly generating websites from handwritten text and identifying humorous elements within images. These features are rarely observed in previous vision-language models. However, the technical details behind GPT-4 continue to remain undisclosed. We believe that the enhanced multi-modal generation capabilities of GPT-4 stem from the utilization of sophisticated large language models (LLM). To examine this phenomenon, we present MiniGPT-4, which aligns a frozen visual encoder with a frozen advanced LLM, Vicuna, using one projection layer. Our work, for the first time, uncovers that properly aligning the visual features with an advanced large language model can possess numerous advanced multi-modal abilities demonstrated by GPT-4, such as detailed image description generation and website creation from hand-drawn drafts. Furthermore, we also observe other emerging capabilities in MiniGPT-4, including writing stories and poems inspired by given images, teaching users how to cook based on food photos, and so on. In our experiment, we found that the model trained on short image caption pairs could produce unnatural language outputs (e.g., repetition and fragmentation). To address this problem, we curate a detailed image description dataset in the second stage to finetune the model, which consequently improves the model's generation reliability and overall usability. Our code, pre-trained model, and collected dataset are available at https://minigpt-4.github.io/.
Visual reasoning is a long-term goal of vision research. In the last decade, several works have attempted to apply deep neural networks (DNNs) to the task of learning visual relations from images, with modest results in terms of the generalization of the relations learned. In recent years, several innovations in DNNs have been developed in order to enable learning abstract relation from images. In this work, we systematically evaluate a series of DNNs that integrate mechanism such as slot attention, recurrently guided attention, and external memory, in the simplest possible visual reasoning task: deciding whether two objects are the same or different. We found that, although some models performed better than others in generalizing the same-different relation to specific types of images, no model was able to generalize this relation across the board. We conclude that abstract visual reasoning remains largely an unresolved challenge for DNNs.
The transferability of adversarial examples is a crucial aspect of evaluating the robustness of deep learning systems, particularly in black-box scenarios. Although several methods have been proposed to enhance cross-model transferability, little attention has been paid to the transferability of adversarial examples across different tasks. This issue has become increasingly relevant with the emergence of foundational multi-task AI systems such as Visual ChatGPT, rendering the utility of adversarial samples generated by a single task relatively limited. Furthermore, these systems often entail inferential functions beyond mere recognition-like tasks. To address this gap, we propose a novel Visual Relation-based cross-task Adversarial Patch generation method called VRAP, which aims to evaluate the robustness of various visual tasks, especially those involving visual reasoning, such as Visual Question Answering and Image Captioning. VRAP employs scene graphs to combine object recognition-based deception with predicate-based relations elimination, thereby disrupting the visual reasoning information shared among inferential tasks. Our extensive experiments demonstrate that VRAP significantly surpasses previous methods in terms of black-box transferability across diverse visual reasoning tasks.
Visual and linguistic pre-training aims to learn vision and language representations together, which can be transferred to visual-linguistic downstream tasks. However, there exists semantic confusion between language and vision during the pre-training stage. Moreover, current pre-trained models tend to take lots of computation resources for fine-tuning when transferred to downstream tasks. In this work, we present a simple but effective approach for learning Contrastive and Adaptive representations of Vision and Language, namely CAVL. Specifically, we introduce a pair-wise contrastive loss to learn alignments between the whole sentence and each image in the same batch during the pre-training process. At the fine-tuning stage, we introduce two lightweight adaptation networks to reduce model parameters and increase training speed for saving computation resources. We evaluate our CAVL on six main downstream tasks, including Visual Question Answering (VQA), Visual Commonsense Reasoning (VCR), Natural Language for Visual Reasoning (NLVR), Region-to-Phrase Grounding (RPG), Text-to-Image Retrieval (TIR), and Zero-shot Text-to-Image Retrieval (ZS-TIR). Compared to baselines, we achieve superior performance and reduce the fine-tuning time by a large margin (in particular, 76.17%). Extensive experiments and ablation studies demonstrate the efficiency of contrastive pre-training and adaptive fine-tuning proposed in our CAVL.
Why do explainable AI (XAI) explanations in radiology, despite their promise of transparency, still fail to gain human trust? Current XAI approaches provide justification for predictions, however, these do not meet practitioners' needs. These XAI explanations lack intuitive coverage of the evidentiary basis for a given classification, posing a significant barrier to adoption. We posit that XAI explanations that mirror human processes of reasoning and justification with evidence may be more useful and trustworthy than traditional visual explanations like heat maps. Using a radiology case study, we demonstrate how radiology practitioners get other practitioners to see a diagnostic conclusion's validity. Machine-learned classifications lack this evidentiary grounding and consequently fail to elicit trust and adoption by potential users. Insights from this study may generalize to guiding principles for human-centered explanation design based on human reasoning and justification of evidence.
Images are a convenient way to specify which particular object instance an embodied agent should navigate to. Solving this task requires semantic visual reasoning and exploration of unknown environments. We present a system that can perform this task in both simulation and the real world. Our modular method solves sub-tasks of exploration, goal instance re-identification, goal localization, and local navigation. We re-identify the goal instance in egocentric vision using feature-matching and localize the goal instance by projecting matched features to a map. Each sub-task is solved using off-the-shelf components requiring zero fine-tuning. On the HM3D InstanceImageNav benchmark, this system outperforms a baseline end-to-end RL policy 7x and a state-of-the-art ImageNav model 2.3x (56% vs 25% success). We deploy this system to a mobile robot platform and demonstrate effective real-world performance, achieving an 88% success rate across a home and an office environment.
Large-scale pre-trained Vision & Language (VL) models have shown remarkable performance in many applications, enabling replacing a fixed set of supported classes with zero-shot open vocabulary reasoning over (almost arbitrary) natural language prompts. However, recent works have uncovered a fundamental weakness of these models. For example, their difficulty to understand Visual Language Concepts (VLC) that go 'beyond nouns' such as the meaning of non-object words (e.g., attributes, actions, relations, states, etc.), or difficulty in performing compositional reasoning such as understanding the significance of the order of the words in a sentence. In this work, we investigate to which extent purely synthetic data could be leveraged to teach these models to overcome such shortcomings without compromising their zero-shot capabilities. We contribute Synthetic Visual Concepts (SyViC) - a million-scale synthetic dataset and data generation codebase allowing to generate additional suitable data to improve VLC understanding and compositional reasoning of VL models. Additionally, we propose a general VL finetuning strategy for effectively leveraging SyViC towards achieving these improvements. Our extensive experiments and ablations on VL-Checklist, Winoground, and ARO benchmarks demonstrate that it is possible to adapt strong pre-trained VL models with synthetic data significantly enhancing their VLC understanding (e.g. by 9.9% on ARO and 4.3% on VL-Checklist) with under 1% drop in their zero-shot accuracy.
The recent wave of large-scale text-to-image diffusion models has dramatically increased our text-based image generation abilities. These models can generate realistic images for a staggering variety of prompts and exhibit impressive compositional generalization abilities. Almost all use cases thus far have solely focused on sampling; however, diffusion models can also provide conditional density estimates, which are useful for tasks beyond image generation. In this paper, we show that the density estimates from large-scale text-to-image diffusion models like Stable Diffusion can be leveraged to perform zero-shot classification without any additional training. Our generative approach to classification, which we call Diffusion Classifier, attains strong results on a variety of benchmarks and outperforms alternative methods of extracting knowledge from diffusion models. Although a gap remains between generative and discriminative approaches on zero-shot recognition tasks, our diffusion-based approach has significantly stronger multimodal compositional reasoning ability than competing discriminative approaches. Finally, we use Diffusion Classifier to extract standard classifiers from class-conditional diffusion models trained on ImageNet. Our models achieve strong classification performance using only weak augmentations and exhibit qualitatively better "effective robustness" to distribution shift. Overall, our results are a step toward using generative over discriminative models for downstream tasks. Results and visualizations at https://diffusion-classifier.github.io/