Insight-V: Exploring Long-Chain Visual Reasoning with Multimodal Large Language Models

Large Language Models (LLMs) demonstrate enhanced capabilities and reliability by reasoning more, evolving from Chain-of-Thought prompting to product-level solutions like OpenAI o1. Despite various efforts to improve LLM reasoning, high-quality long-chain reasoning data and optimized training pipelines still remain inadequately explored in vision-language tasks. In this paper, we present Insight-V, an early effort to 1) scalably produce long and robust reasoning data for complex multi-modal tasks, and 2) an effective training pipeline to enhance the reasoning capabilities of multi-modal large language models (MLLMs). Specifically, to create long and structured reasoning data without human labor, we design a two-step pipeline with a progressive strategy to generate sufficiently long and diverse reasoning paths and a multi-granularity assessment method to ensure data quality. We observe that directly supervising MLLMs with such long and complex reasoning data will not yield ideal reasoning ability. To tackle this problem, we design a multi-agent system consisting of a reasoning agent dedicated to performing long-chain reasoning and a summary agent trained to judge and summarize reasoning results. We further incorporate an iterative DPO algorithm to enhance the reasoning agent's generation stability and quality. Based on the popular LLaVA-NeXT model and our stronger base MLLM, we demonstrate significant performance gains across challenging multi-modal benchmarks requiring visual reasoning. Benefiting from our multi-agent system, Insight-V can also easily maintain or improve performance on perception-focused multi-modal tasks.

Coarse Correspondence Elicit 3D Spacetime Understanding in Multimodal Language Model

Multimodal language models (MLLMs) are increasingly being implemented in real-world environments, necessitating their ability to interpret 3D spaces and comprehend temporal dynamics. Despite their potential, current top models within our community still fall short in adequately understanding spatial and temporal dimensions. We introduce Coarse Correspondence, a simple, training-free, effective, and general-purpose visual prompting method to elicit 3D and temporal understanding in multimodal LLMs. Our method uses a lightweight tracking model to find object correspondences between frames in a video or between sets of image viewpoints. It selects the most frequent object instances and visualizes them with markers with unique IDs in the image. With this simple approach, we achieve state-of-the-art results on 3D understanding benchmarks including ScanQA (+20.5\%) and a subset of OpenEQA (+9.7\%), and on long-form video benchmarks such as EgoSchema (+6.0\%). We also curate a small diagnostic dataset to evaluate whether MLLMs can reason about space from a described viewpoint other than the camera viewpoint. Again, Coarse Correspondence improves spatial perspective-taking abilities but we highlight that MLLMs struggle with this task. Together, we demonstrate that our simple prompting method can significantly aid downstream tasks that require 3D or temporal reasoning.

Efficient Inference of Vision Instruction-Following Models with Elastic Cache

In the field of instruction-following large vision-language models (LVLMs), the efficient deployment of these models faces challenges, notably due to the high memory demands of their key-value (KV) caches. Conventional cache management strategies for LLMs focus on cache eviction, which often fails to address the specific needs of multimodal instruction-following models. Recognizing this gap, in this paper, we introduce Elastic Cache, a novel approach that benefits from applying distinct acceleration methods for instruction encoding and output generation stages. We investigate the metrics of importance in different stages and propose an importance-driven cache merging strategy to prune redundancy caches. Instead of discarding less important caches, our strategy identifies important key/value vectors as anchor points. Surrounding less important caches are then merged with these anchors, enhancing the preservation of contextual information in the KV caches while yielding an arbitrary acceleration ratio. For instruction encoding, we utilize the frequency to evaluate the importance of caches. Regarding output generation, we prioritize tokens based on their distance with an offset, by which both the initial and most recent tokens are retained. Results on a range of LVLMs demonstrate that Elastic Cache not only boosts efficiency but also notably outperforms existing pruning methods in language generation across various tasks. Code is available at https://github.com/liuzuyan/ElasticCache

Chain-of-Spot: Interactive Reasoning Improves Large Vision-Language Models

In the realm of vision-language understanding, the proficiency of models in interpreting and reasoning over visual content has become a cornerstone for numerous applications. However, it is challenging for the visual encoder in Large Vision-Language Models (LVLMs) to extract useful features tailored to questions that aid the language model's response. Furthermore, a common practice among existing LVLMs is to utilize lower-resolution images, which restricts the ability for visual recognition. Our work introduces the Chain-of-Spot (CoS) method, which we describe as Interactive Reasoning, a novel approach that enhances feature extraction by focusing on key regions of interest (ROI) within the image, corresponding to the posed questions or instructions. This technique allows LVLMs to access more detailed visual information without altering the original image resolution, thereby offering multi-granularity image features. By integrating Chain-of-Spot with instruct-following LLaVA-1.5 models, the process of image reasoning consistently improves performance across a wide range of multimodal datasets and benchmarks without bells and whistles and achieves new state-of-the-art results. Our empirical findings demonstrate a significant improvement in LVLMs' ability to understand and reason about visual content, paving the way for more sophisticated visual instruction-following applications. Code and models are available at https://github.com/dongyh20/Chain-of-Spot

NSM4D: Neural Scene Model Based Online 4D Point Cloud Sequence Understanding

Understanding 4D point cloud sequences online is of significant practical value in various scenarios such as VR/AR, robotics, and autonomous driving. The key goal is to continuously analyze the geometry and dynamics of a 3D scene as unstructured and redundant point cloud sequences arrive. And the main challenge is to effectively model the long-term history while keeping computational costs manageable. To tackle these challenges, we introduce a generic online 4D perception paradigm called NSM4D. NSM4D serves as a plug-and-play strategy that can be adapted to existing 4D backbones, significantly enhancing their online perception capabilities for both indoor and outdoor scenarios. To efficiently capture the redundant 4D history, we propose a neural scene model that factorizes geometry and motion information by constructing geometry tokens separately storing geometry and motion features. Exploiting the history becomes as straightforward as querying the neural scene model. As the sequence progresses, the neural scene model dynamically deforms to align with new observations, effectively providing the historical context and updating itself with the new observations. By employing token representation, NSM4D also exhibits robustness to low-level sensor noise and maintains a compact size through a geometric sampling scheme. We integrate NSM4D with state-of-the-art 4D perception backbones, demonstrating significant improvements on various online perception benchmarks in indoor and outdoor settings. Notably, we achieve a 9.6% accuracy improvement for HOI4D online action segmentation and a 3.4% mIoU improvement for SemanticKITTI online semantic segmentation. Furthermore, we show that NSM4D inherently offers excellent scalability to longer sequences beyond the training set, which is crucial for real-world applications.

Octopus: Embodied Vision-Language Programmer from Environmental Feedback

Large vision-language models (VLMs) have achieved substantial progress in multimodal perception and reasoning. Furthermore, when seamlessly integrated into an embodied agent, it signifies a crucial stride towards the creation of autonomous and context-aware systems capable of formulating plans and executing commands with precision. In this paper, we introduce Octopus, a novel VLM designed to proficiently decipher an agent's vision and textual task objectives and to formulate intricate action sequences and generate executable code. Our design allows the agent to adeptly handle a wide spectrum of tasks, ranging from mundane daily chores in simulators to sophisticated interactions in complex video games. Octopus is trained by leveraging GPT-4 to control an explorative agent to generate training data, i.e., action blueprints and the corresponding executable code, within our experimental environment called OctoVerse. We also collect the feedback that allows the enhanced training scheme of Reinforcement Learning with Environmental Feedback (RLEF). Through a series of experiments, we illuminate Octopus's functionality and present compelling results, and the proposed RLEF turns out to refine the agent's decision-making. By open-sourcing our model architecture, simulator, and dataset, we aspire to ignite further innovation and foster collaborative applications within the broader embodied AI community.

Complete-to-Partial 4D Distillation for Self-Supervised Point Cloud Sequence Representation Learning

Recent work on 4D point cloud sequences has attracted a lot of attention. However, obtaining exhaustively labeled 4D datasets is often very expensive and laborious, so it is especially important to investigate how to utilize raw unlabeled data. However, most existing self-supervised point cloud representation learning methods only consider geometry from a static snapshot omitting the fact that sequential observations of dynamic scenes could reveal more comprehensive geometric details. And the video representation learning frameworks mostly model motion as image space flows, let alone being 3D-geometric-aware. To overcome such issues, this paper proposes a new 4D self-supervised pre-training method called Complete-to-Partial 4D Distillation. Our key idea is to formulate 4D self-supervised representation learning as a teacher-student knowledge distillation framework and let the student learn useful 4D representations with the guidance of the teacher. Experiments show that this approach significantly outperforms previous pre-training approaches on a wide range of 4D point cloud sequence understanding tasks including indoor and outdoor scenarios.

ImageTBAD: A 3D Computed Tomography Angiography Image Dataset for Automatic Segmentation of Type-B Aortic Dissection

Type-B Aortic Dissection (TBAD) is one of the most serious cardiovascular events characterized by a growing yearly incidence,and the severity of disease prognosis. Currently, computed tomography angiography (CTA) has been widely adopted for the diagnosis and prognosis of TBAD. Accurate segmentation of true lumen (TL), false lumen (FL), and false lumen thrombus (FLT) in CTA are crucial for the precise quantification of anatomical features. However, existing works only focus on only TL and FL without considering FLT. In this paper, we propose ImageTBAD, the first 3D computed tomography angiography (CTA) image dataset of TBAD with annotation of TL, FL, and FLT. The proposed dataset contains 100 TBAD CTA images, which is of decent size compared with existing medical imaging datasets. As FLT can appear almost anywhere along the aorta with irregular shapes, segmentation of FLT presents a wide class of segmentation problems where targets exist in a variety of positions with irregular shapes. We further propose a baseline method for automatic segmentation of TBAD. Results show that the baseline method can achieve comparable results with existing works on aorta and TL segmentation. However, the segmentation accuracy of FLT is only 52%, which leaves large room for improvement and also shows the challenge of our dataset. To facilitate further research on this challenging problem, our dataset and codes are released to the public.

Multi-Cycle-Consistent Adversarial Networks for Edge Denoising of Computed Tomography Images

As one of the most commonly ordered imaging tests, computed tomography (CT) scan comes with inevitable radiation exposure that increases the cancer risk to patients. However, CT image quality is directly related to radiation dose, thus it is desirable to obtain high-quality CT images with as little dose as possible. CT image denoising tries to obtain high dose like high-quality CT images (domain X) from low dose low-quality CTimages (domain Y), which can be treated as an image-to-image translation task where the goal is to learn the transform between a source domain X (noisy images) and a target domain Y (clean images). In this paper, we propose a multi-cycle-consistent adversarial network (MCCAN) that builds intermediate domains and enforces both local and global cycle-consistency for edge denoising of CT images. The global cycle-consistency couples all generators together to model the whole denoising process, while the local cycle-consistency imposes effective supervision on the process between adjacent domains. Experiments show that both local and global cycle-consistency are important for the success of MCCAN, which outperformsCCADN in terms of denoising quality with slightly less computation resource consumption.

ImageCHD: A 3D Computed Tomography Image Dataset for Classification of Congenital Heart Disease

Congenital heart disease (CHD) is the most common type of birth defect, which occurs 1 in every 110 births in the United States. CHD usually comes with severe variations in heart structure and great artery connections that can be classified into many types. Thus highly specialized domain knowledge and the time-consuming human process is needed to analyze the associated medical images. On the other hand, due to the complexity of CHD and the lack of dataset, little has been explored on the automatic diagnosis (classification) of CHDs. In this paper, we present ImageCHD, the first medical image dataset for CHD classification. ImageCHD contains 110 3D Computed Tomography (CT) images covering most types of CHD, which is of decent size Classification of CHDs requires the identification of large structural changes without any local tissue changes, with limited data. It is an example of a larger class of problems that are quite difficult for current machine-learning-based vision methods to solve. To demonstrate this, we further present a baseline framework for the automatic classification of CHD, based on a state-of-the-art CHD segmentation method. Experimental results show that the baseline framework can only achieve a classification accuracy of 82.0\% under a selective prediction scheme with 88.4\% coverage, leaving big room for further improvement. We hope that ImageCHD can stimulate further research and lead to innovative and generic solutions that would have an impact in multiple domains. Our dataset is released to the public compared with existing medical imaging datasets.