Abstract:Chart understanding plays a pivotal role when applying Multimodal Large Language Models (MLLMs) to real-world tasks such as analyzing scientific papers or financial reports. However, existing datasets often focus on oversimplified and homogeneous charts with template-based questions, leading to an over-optimistic measure of progress. We demonstrate that although open-source models can appear to outperform strong proprietary models on these benchmarks, a simple stress test with slightly different charts or questions can deteriorate performance by up to 34.5%. In this work, we propose CharXiv, a comprehensive evaluation suite involving 2,323 natural, challenging, and diverse charts from arXiv papers. CharXiv includes two types of questions: 1) descriptive questions about examining basic chart elements and 2) reasoning questions that require synthesizing information across complex visual elements in the chart. To ensure quality, all charts and questions are handpicked, curated, and verified by human experts. Our results reveal a substantial, previously underestimated gap between the reasoning skills of the strongest proprietary model (i.e., GPT-4o), which achieves 47.1% accuracy, and the strongest open-source model (i.e., InternVL Chat V1.5), which achieves 29.2%. All models lag far behind human performance of 80.5%, underscoring weaknesses in the chart understanding capabilities of existing MLLMs. We hope CharXiv facilitates future research on MLLM chart understanding by providing a more realistic and faithful measure of progress. Project page and leaderboard: https://charxiv.github.io/
Abstract:Recent studies show that image and video generation models can be prompted to reproduce copyrighted content from their training data, raising serious legal concerns around copyright infringement. Copyrighted characters, in particular, pose a difficult challenge for image generation services, with at least one lawsuit already awarding damages based on the generation of these characters. Yet, little research has empirically examined this issue. We conduct a systematic evaluation to fill this gap. First, we build CopyCat, an evaluation suite consisting of diverse copyrighted characters and a novel evaluation pipeline. Our evaluation considers both the detection of similarity to copyrighted characters and generated image's consistency with user input. Our evaluation systematically shows that both image and video generation models can still generate characters even if characters' names are not explicitly mentioned in the prompt, sometimes with only two generic keywords (e.g., prompting with "videogame, plumber" consistently generates Nintendo's Mario character). We then introduce techniques to semi-automatically identify such keywords or descriptions that trigger character generation. Using our evaluation suite, we study runtime mitigation strategies, including both existing methods and new strategies we propose. Our findings reveal that commonly employed strategies, such as prompt rewriting in the DALL-E system, are not sufficient as standalone guardrails. These strategies must be coupled with other approaches, like negative prompting, to effectively reduce the unintended generation of copyrighted characters. Our work provides empirical grounding to the discussion of copyright mitigation strategies and offers actionable insights for model deployers actively implementing them.
Abstract:Humans can imagine goal states during planning and perform actions to match those goals. In this work, we propose Imagination Policy, a novel multi-task key-frame policy network for solving high-precision pick and place tasks. Instead of learning actions directly, Imagination Policy generates point clouds to imagine desired states which are then translated to actions using rigid action estimation. This transforms action inference into a local generative task. We leverage pick and place symmetries underlying the tasks in the generation process and achieve extremely high sample efficiency and generalizability to unseen configurations. Finally, we demonstrate state-of-the-art performance across various tasks on the RLbench benchmark compared with several strong baselines.
Abstract:Large Multimodal Models (LMMs) have shown remarkable capabilities across a variety of tasks (e.g., image captioning, visual question answering). While broad, their knowledge remains generic (e.g., recognizing a dog), and they are unable to handle personalized subjects (e.g., recognizing a user's pet dog). Human reasoning, in contrast, typically operates within the context of specific subjects in our surroundings. For example, one might ask, "What should I buy for my dog's birthday?"; as opposed to a generic inquiry about "What should I buy for a dog's birthday?". Similarly, when looking at a friend's image, the interest lies in seeing their activities (e.g., "my friend is holding a cat"), rather than merely observing generic human actions (e.g., "a man is holding a cat"). In this paper, we introduce the novel task of personalizing LMMs, so that they can have conversations about a specific subject. We propose Yo'LLaVA, which learns to embed a personalized subject into a set of latent tokens given a handful of example images of the subject. Our qualitative and quantitative analyses reveal that Yo'LLaVA can learn the concept more efficiently using fewer tokens and more effectively encode the visual attributes compared to strong prompting baselines (e.g., LLaVA).
Abstract:While following different technical routes, both low-rank and orthogonal adaptation techniques can efficiently adapt large-scale pre-training models in specific tasks or domains based on a small piece of trainable parameters. In this study, we bridge the gap between these two techniques, proposing a simple but effective adaptation method based on Householder reflections. Given a pre-trained model, our method fine-tunes its layers by multiplying each frozen weight matrix with an orthogonal matrix constructed by a chain of learnable Householder reflections (HRs). This HR-based orthogonal fine-tuning is equivalent to an adaptive low-rank adaptation. Moreover, we show that the orthogonality of the reflection planes corresponding to the HRs impacts the model capacity and regularity. The analysis motivates us to regularize the orthogonality of the HRs, leading to different implementations of the proposed Householder reflection adaptation (HRA) method. Compared with state-of-the-art methods, HRA achieves superior performance with fewer learnable parameters when adapting large language models and conditional image generators. The code is available at https://github.com/DaShenZi721/HRA
Abstract:In the evolution towards the forthcoming era of sixth-generation (6G) mobile communication systems characterized by ubiquitous intelligence, integrated sensing and communication (ISAC) is in a phase of burgeoning development. However, the capabilities of communication and sensing within single frequency band fall short of meeting the escalating demands. To this end, this paper introduces a carrier aggregation (CA)- enabled multi-input multi-output orthogonal frequency division multiplexing (MIMO-OFDM) ISAC system fusing the sensing data on high and low-frequency bands by symbol-level fusion for ultimate communication experience and high-accuracy sensing. The challenges in sensing signal processing introduced by CA include the initial phase misalignment of the echo signals on high and low-frequency bands due to attenuation and radar cross section, and the fusion of the sensing data on high and lowfrequency bands with different physical-layer parameters. To this end, the sensing signal processing is decomposed into two stages. In the first stage, the problem of initial phase misalignment of the echo signals on high and low-frequency bands is solved by the angle compensation, space-domain diversity and vector crosscorrelation operations. In the second stage, this paper realizes symbol-level fusion of the sensing data on high and low-frequency bands through sensing vector rearrangement and cyclic prefix adjustment operations, thereby obtaining high-precision sensing performance. Then, the closed-form communication mutual information (MI) and sensing Cramer-Rao lower bound (CRLB) for the proposed ISAC system are derived to explore the theoretical performance bound with CA. Simulation results validate the feasibility and superiority of the proposed ISAC system.
Abstract:Addressing the communication and sensing demands of sixth-generation (6G) mobile communication system, integrated sensing and communication (ISAC) has garnered traction in academia and industry. With the sensing limitation of single base station (BS), multi-BS cooperative sensing is regarded as a promising solution. The coexistence and overlapped coverage of macro BS (MBS) and micro BS (MiBS) are common in the development of 6G, making the cooperative sensing between MBS and MiBS feasible. Since MBS and MiBS work in low and high frequency bands, respectively, the challenges of MBS and MiBS cooperative sensing lie in the fusion method of the sensing information in high and low-frequency bands. To this end, this paper introduces a symbol-level fusion method and a grid-based three-dimensional discrete Fourier transform (3D-GDFT) algorithm to achieve precise localization of multiple targets with limited resources. Simulation results demonstrate that the proposed MBS and MiBS cooperative sensing scheme outperforms traditional single BS (MBS/MiBS) sensing scheme, showcasing superior sensing performance
Abstract:As a significant step for human face modeling, editing, and generation, face landmarking aims at extracting facial keypoints from images. A generalizable face landmarker is required in practice because real-world facial images, e.g., the avatars in animations and games, are often stylized in various ways. However, achieving generalizable face landmarking is challenging due to the diversity of facial styles and the scarcity of labeled stylized faces. In this study, we propose a simple but effective paradigm to learn a generalizable face landmarker based on labeled real human faces and unlabeled stylized faces. Our method learns the face landmarker as the key module of a conditional face warper. Given a pair of real and stylized facial images, the conditional face warper predicts a warping field from the real face to the stylized one, in which the face landmarker predicts the ending points of the warping field and provides us with high-quality pseudo landmarks for the corresponding stylized facial images. Applying an alternating optimization strategy, we learn the face landmarker to minimize $i)$ the discrepancy between the stylized faces and the warped real ones and $ii)$ the prediction errors of both real and pseudo landmarks. Experiments on various datasets show that our method outperforms existing state-of-the-art domain adaptation methods in face landmarking tasks, leading to a face landmarker with better generalizability. Code is available at https://plustwo0.github.io/project-face-landmarker}{https://plustwo0.github.io/project-face-landmarker.
Abstract:Aerial vehicles equipped with manipulators can serve contact-based industrial applications, where fundamental tasks like drilling and grinding often necessitate aerial platforms to handle heavy tools. Industrial environments often involve non-horizontal surfaces. Existing aerial manipulation platforms based on multirotors typically feature a fixed CoM (Center of Mass) within the rotor-defined area, leading to a considerable moment arm between the EE (End-Effector) tip and the CoM for operations on such surfaces. Carrying heavy tools at the EE tip of the manipulator with an extended moment arm can lead to system instability and potential damage to the servo actuators used in the manipulator. To tackle this issue, we present a novel aerial vehicle tailored for handling heavy tools on non-horizontal surfaces. In this work, we provide the platform's system design, modeling, and control strategies. This platform can carry heavy manipulators within the rotor-defined area during free flight. During interactions, the manipulator can shift towards the work surface outside the rotor-defined area, resulting in a displaced CoM location with a significantly shorter moment arm. Furthermore, we propose a method for automatically determining the manipulator's position to reach the maximum CoM displacement towards the work surface. Our proposed concepts are validated through simulations that closely capture the developed physical prototype of the platform.
Abstract:With the mobile communication system evolving into 6th-generation (6G), the Internet of Everything (IoE) is becoming reality, which connects human, big data and intelligent machines to support the intelligent decision making, reconfiguring the traditional industries and human life. The applications of IoE require not only pure communication capability, but also high-accuracy and large-scale sensing capability. With the emerging integrated sensing and communication (ISAC) technique, exploiting the mobile communication system with multi-domain resources, multiple network elements, and large-scale infrastructures to realize cooperative sensing is a crucial approach to satisfy the requirements of high-accuracy and large-scale sensing in IoE. In this article, the deep cooperation in ISAC system including three perspectives is investigated. In the microscopic perspective, namely, within a single node, the cooperation at the resource-level is performed to improve sensing accuracy by fusing the sensing information carried in the time-frequency-space-code multi-domain resources. In the mesoscopic perspective, the sensing accuracy could be improved through the cooperation of multiple nodes including Base Station (BS), User Equipment (UE), and Reconfigurable Intelligence Surface (RIS), etc. In the macroscopic perspective, the massive number of infrastructures from the same operator or different operators could perform cooperative sensing to extend the sensing coverage and improve the sensing continuity. This article may provide a deep and comprehensive view on the cooperative sensing in ISAC system to enhance the performance of sensing, supporting the applications of IoE.