arXiv subjects

Bohao Li

Publications and source records attributed to Bohao Li.

33 records · Page 2Linked to original sources

arXiv2024

VidEgoThink: Assessing Egocentric Video Understanding Capabilities for Embodied AI

Recent advancements in Multi-modal Large Language Models (MLLMs) have opened new avenues for applications in Embodied AI. Building on previous work, EgoThink, we introduce VidEgoThink, a comprehensive benchmark for evaluating egocentric video understanding capabilities. To bridge the gap between MLLMs and low-level control in Embodied AI, we design four key interrelated tasks: video question-answering, hierarchy planning, visual grounding and reward modeling. To minimize manual annotation costs, we develop an automatic data generation pipeline based on the Ego4D dataset, leveraging the prior knowledge and multimodal capabilities of GPT-4o. Three human annotators then filter the generated data to ensure diversity and quality, resulting in the VidEgoThink benchmark. We conduct extensive experiments with three types of models: API-based MLLMs, open-source image-based MLLMs, and open-source video-based MLLMs. Experimental results indicate that all MLLMs, including GPT-4o, perform poorly across all tasks related to egocentric video understanding. These findings suggest that foundation models still require significant advancements to be effectively applied to first-person scenarios in Embodied AI. In conclusion, VidEgoThink reflects a research trend towards employing MLLMs for egocentric vision, akin to human capabilities, enabling active observation and interaction in the complex real-world environments.

cs.CV↗

arXiv2024

EgoPlan-Bench: Benchmarking Multimodal Large Language Models for Human-Level Planning

The pursuit of artificial general intelligence (AGI) has been accelerated by Multimodal Large Language Models (MLLMs), which exhibit superior reasoning, generalization capabilities, and proficiency in processing multimodal inputs. A crucial milestone in the evolution of AGI is the attainment of human-level planning, a fundamental ability for making informed decisions in complex environments, and solving a wide range of real-world problems. Despite the impressive advancements in MLLMs, a question remains: How far are current MLLMs from achieving human-level planning? To shed light on this question, we introduce EgoPlan-Bench, a comprehensive benchmark to evaluate the planning abilities of MLLMs in real-world scenarios from an egocentric perspective, mirroring human perception. EgoPlan-Bench emphasizes the evaluation of planning capabilities of MLLMs, featuring realistic tasks, diverse action plans, and intricate visual observations. Our rigorous evaluation of a wide range of MLLMs reveals that EgoPlan-Bench poses significant challenges, highlighting a substantial scope for improvement in MLLMs to achieve human-level task planning. To facilitate this advancement, we further present EgoPlan-IT, a specialized instruction-tuning dataset that effectively enhances model performance on EgoPlan-Bench. We have made all codes, data, and a maintained benchmark leaderboard available to advance future research.

cs.CV↗

arXiv2024

Tunable superconductivity in electron- and hole-doped Bernal bilayer graphene

Graphene-based, high quality two-dimensional electronic systems have emerged as a highly tunable platform for studying superconductivity. Specifically, superconductivity has been observed in both electron-doped and hole-doped twisted graphene moire systems, whereas in crystalline graphene systems, superconductivity has so far only been observed in hole-doped rhombohedral trilayer and hole-doped Bernal bilayer graphene (BBG). Recently, enhanced superconductivity has been demonstrated in BBG due to the proximity with a monolayer WSe2. Here, we report the observation of superconductivity and a series of flavor-symmetry-breaking phases in both electron- and hole-doped BBG/WSe2 device by electrostatic doping. The strength of the observed superconductivity is tunable by applied vertical electric fields. The maximum Berezinskii-Kosterlitz-Thouless (BKT) transition temperature for the electron- and hole-doped superconductivity is about 210 mK and 400 mK, respectively. Superconductivities emerge only when applied electric fields drive BBG electron or hole wavefunctions toward the WSe2 layer, underscoring the importance of the WSe2 layer in the observed superconductivity. We find the hole-doped superconductivity violates the Pauli paramagnetic limit, consistent with an Ising-like superconductor. In contrast, the electron-doped superconductivity obeys the Pauli limit, even though the proximity induced Ising spin-orbit coupling is also notable in the conduction band. Our findings highlight the rich physics associated with the conduction band in BBG, paving the way for further studies into the superconducting mechanisms of crystalline graphene and the development of novel superconductor devices based on BBG.

cond-mat.supr-con↗

arXiv2024

SEED-Bench-2-Plus: Benchmarking Multimodal Large Language Models with Text-Rich Visual Comprehension

Comprehending text-rich visual content is paramount for the practical application of Multimodal Large Language Models (MLLMs), since text-rich scenarios are ubiquitous in the real world, which are characterized by the presence of extensive texts embedded within images. Recently, the advent of MLLMs with impressive versatility has raised the bar for what we can expect from MLLMs. However, their proficiency in text-rich scenarios has yet to be comprehensively and objectively assessed, since current MLLM benchmarks primarily focus on evaluating general visual comprehension. In this work, we introduce SEED-Bench-2-Plus, a benchmark specifically designed for evaluating \textbf{text-rich visual comprehension} of MLLMs. Our benchmark comprises 2.3K multiple-choice questions with precise human annotations, spanning three broad categories: Charts, Maps, and Webs, each of which covers a wide spectrum of text-rich scenarios in the real world. These categories, due to their inherent complexity and diversity, effectively simulate real-world text-rich environments. We further conduct a thorough evaluation involving 34 prominent MLLMs (including GPT-4V, Gemini-Pro-Vision and Claude-3-Opus) and emphasize the current limitations of MLLMs in text-rich visual comprehension. We hope that our work can serve as a valuable addition to existing MLLM benchmarks, providing insightful observations and inspiring further research in the area of text-rich visual comprehension with MLLMs. The dataset and evaluation code can be accessed at https://github.com/AILab-CVC/SEED-Bench.

cs.CV↗

arXiv2024

Electrically tuned topology and magnetism in twisted bilayer MoTe$_2$ at $ν_h=1$

We present a theoretical study of an interaction-driven quantum phase diagram of twisted bilayer MoTe$_2$ at hole filling factor $ν_h=1$ as a function of twist angle $θ$ and layer potential difference $V_z$, where $V_z$ is generated by an applied out-of-plane electric field. At $V_z=0$, the phase diagram includes quantum anomalous Hall insulators in the intermediate $θ$ regime and topologically trivial multiferroic states with coexisting ferroelectricity and magnetism in both small and large $θ$ regimes. There can be two transitions from the quantum anomalous Hall insulator phase to topologically trivial out-of-plane ferromagnetic phase, and finally to in-plane 120$^\circ$ antiferromagnetic phase as $|V_z|$ increases, or a single transition without the intervening ferromagnetic phase. We show explicitly that the spin vector chirality of various 120$^\circ$ antiferromagnetic states can be electrically switched. We discuss the connection between the experimentally measured Curie-Weiss temperature and the low-temperature magnetic order based on an effective Heisenberg model with magnetic anisotropy.

cond-mat.mes-hall↗

arXiv2023

SEED-Bench-2: Benchmarking Multimodal Large Language Models

Multimodal large language models (MLLMs), building upon the foundation of powerful large language models (LLMs), have recently demonstrated exceptional capabilities in generating not only texts but also images given interleaved multimodal inputs (acting like a combination of GPT-4V and DALL-E 3). However, existing MLLM benchmarks remain limited to assessing only models' comprehension ability of single image-text inputs, failing to keep up with the strides made in MLLMs. A comprehensive benchmark is imperative for investigating the progress and uncovering the limitations of current MLLMs. In this work, we categorize the capabilities of MLLMs into hierarchical levels from $L_0$ to $L_4$ based on the modalities they can accept and generate, and propose SEED-Bench-2, a comprehensive benchmark that evaluates the \textbf{hierarchical} capabilities of MLLMs. Specifically, SEED-Bench-2 comprises 24K multiple-choice questions with accurate human annotations, which spans 27 dimensions, including the evaluation of both text and image generation. Multiple-choice questions with groundtruth options derived from human annotation enables an objective and efficient assessment of model performance, eliminating the need for human or GPT intervention during evaluation. We further evaluate the performance of 23 prominent open-source MLLMs and summarize valuable observations. By revealing the limitations of existing MLLMs through extensive evaluations, we aim for SEED-Bench-2 to provide insights that will motivate future research towards the goal of General Artificial Intelligence. Dataset and evaluation code are available at \href{https://github.com/AILab-CVC/SEED-Bench}

cs.CV↗

arXiv2023

Interaction-driven topological phase diagram of twisted bilayer MoTe$_2$

Twisted bilayer MoTe$_2$ is a promising platform to investigate the interplay between band topology and many-body interaction. We present a theoretical study of its interaction-driven quantum phase diagrams based on a three-orbital model, which can be viewed as a generalization of the Kane-Mele-Hubbard model with one additional orbital and long-range Coulomb repulsion. We predict a cascade of phase transitions tuned by the twist angle $θ$. At the hole filling factor $ν=1$ (one hole per moiré unit cell), the ground state can be in the multiferroic phase with coexisting spontaneous layer polarization and magnetism, the quantum anomalous Hall phase, and finally the topologically trivial magnetic phases, as $θ$ increases from $1.5^{\circ}$ to $5^{\circ}$. At $ν=2$, the ground state can have a second-order phase transition between an antiferromagnetic phase and the quantum spin Hall phase as $θ$ passes through a critical value. The dependence of the phase boundaries on model parameters such as the gate-to-sample distance, the dielectric constant, and the moiré potential amplitude is examined. The predicted phase diagrams can guide the search for topological phases in twisted transition metal dichalcogenide homobilayers.

cond-mat.mes-hall↗

arXiv2023

SEED-Bench: Benchmarking Multimodal LLMs with Generative Comprehension

Based on powerful Large Language Models (LLMs), recent generative Multimodal Large Language Models (MLLMs) have gained prominence as a pivotal research area, exhibiting remarkable capability for both comprehension and generation. In this work, we address the evaluation of generative comprehension in MLLMs as a preliminary step towards a comprehensive assessment of generative models, by introducing a benchmark named SEED-Bench. SEED-Bench consists of 19K multiple choice questions with accurate human annotations (x 6 larger than existing benchmarks), which spans 12 evaluation dimensions including the comprehension of both the image and video modality. We develop an advanced pipeline for generating multiple-choice questions that target specific evaluation dimensions, integrating both automatic filtering and manual verification processes. Multiple-choice questions with groundtruth options derived from human annotation enables an objective and efficient assessment of model performance, eliminating the need for human or GPT intervention during evaluation. We further evaluate the performance of 18 models across all 12 dimensions, covering both the spatial and temporal understanding. By revealing the limitations of existing MLLMs through evaluation results, we aim for SEED-Bench to provide insights for motivating future research. We will launch and consistently maintain a leaderboard to provide a platform for the community to assess and investigate model capability.

cs.CL↗

arXiv2023

Prompt, Generate, then Cache: Cascade of Foundation Models makes Strong Few-shot Learners

Visual recognition in low-data regimes requires deep neural networks to learn generalized representations from limited training samples. Recently, CLIP-based methods have shown promising few-shot performance benefited from the contrastive language-image pre-training. We then question, if the more diverse pre-training knowledge can be cascaded to further assist few-shot representation learning. In this paper, we propose CaFo, a Cascade of Foundation models that incorporates diverse prior knowledge of various pre-training paradigms for better few-shot learning. Our CaFo incorporates CLIP's language-contrastive knowledge, DINO's vision-contrastive knowledge, DALL-E's vision-generative knowledge, and GPT-3's language-generative knowledge. Specifically, CaFo works by 'Prompt, Generate, then Cache'. Firstly, we leverage GPT-3 to produce textual inputs for prompting CLIP with rich downstream linguistic semantics. Then, we generate synthetic images via DALL-E to expand the few-shot training data without any manpower. At last, we introduce a learnable cache model to adaptively blend the predictions from CLIP and DINO. By such collaboration, CaFo can fully unleash the potential of different pre-training methods and unify them to perform state-of-the-art for few-shot classification. Code is available at https://github.com/ZrrSkywalker/CaFo.

cs.CV↗

arXiv2022

Proposal Distribution Calibration for Few-Shot Object Detection

Adapting object detectors learned with sufficient supervision to novel classes under low data regimes is charming yet challenging. In few-shot object detection (FSOD), the two-step training paradigm is widely adopted to mitigate the severe sample imbalance, i.e., holistic pre-training on base classes, then partial fine-tuning in a balanced setting with all classes. Since unlabeled instances are suppressed as backgrounds in the base training phase, the learned RPN is prone to produce biased proposals for novel instances, resulting in dramatic performance degradation. Unfortunately, the extreme data scarcity aggravates the proposal distribution bias, hindering the RoI head from evolving toward novel classes. In this paper, we introduce a simple yet effective proposal distribution calibration (PDC) approach to neatly enhance the localization and classification abilities of the RoI head by recycling its localization ability endowed in base training and enriching high-quality positive samples for semantic fine-tuning. Specifically, we sample proposals based on the base proposal statistics to calibrate the distribution bias and impose additional localization and classification losses upon the sampled proposals for fast expanding the base detector to novel classes. Experiments on the commonly used Pascal VOC and MS COCO datasets with explicit state-of-the-art performances justify the efficacy of our PDC for FSOD. Code is available at github.com/Bohao-Lee/PDC.

cs.CV↗

arXiv2022

Collaboration of Pre-trained Models Makes Better Few-shot Learner

Few-shot classification requires deep neural networks to learn generalized representations only from limited training images, which is challenging but significant in low-data regimes. Recently, CLIP-based methods have shown promising few-shot performance benefited from the contrastive language-image pre-training. Based on this point, we question if the large-scale pre-training can alleviate the few-shot data deficiency and also assist the representation learning by the pre-learned knowledge. In this paper, we propose CoMo, a Collaboration of pre-trained Models that incorporates diverse prior knowledge from various pre-training paradigms for better few-shot learning. Our CoMo includes: CLIP's language-contrastive knowledge, DINO's vision-contrastive knowledge, and DALL-E's language-generative knowledge. Specifically, CoMo works in two aspects: few-shot data expansion and diverse knowledge ensemble. For one, we generate synthetic images via zero-shot DALL-E to enrich the few-shot training data without any manpower. For the other, we introduce a learnable Multi-Knowledge Adapter (MK-Adapter) to adaptively blend the predictions from CLIP and DINO. By such collaboration, CoMo can fully unleash the potential of different pre-training methods and unify them to perform state-of-the-art for few-shot classification. We conduct extensive experiments on 11 datasets to demonstrate the superiority and generalization ability of our approach.

cs.CV↗

arXiv2022

Prediction of ferroelectric superconductors with reversible superconducting diode effect

A noncentrosymmetric superconductor can have a superconducting diode effect, where the critical current in opposite directions is different when time-reversal symmetry is also broken. We theoretically propose that a ferroelectric superconductor with coexisting ferroelectricity and superconductivity can support a ferroelectric reversible superconducting diode effect. Through first-principles calculation, we predict that monolayer CuNb$_2$Se$_4$ (i.e., bilayer NbSe$_2$ intercalated with Cu) is such a ferroelectric superconductor, where ferroelectricity controls the layer polarization as well as the sign of spin-orbit coupling induced spin splittings. Because the nonreciprocal effect of the critical current is proportional to the spin splittings, the superconducting diode effect is reversible upon electric switch of ferroelectricity. While we use CuNb$_2$Se$_4$ as a model system, the predicted effect can appear in a class of two-dimensional superconducting bilayers with ferroelectricity induced by interlayer sliding. Our work opens the door to studying the interplay between superconductivity and ferroelectricity in two-dimensional materials.

cond-mat.supr-con↗

arXiv2021

Beyond Max-Margin: Class Margin Equilibrium for Few-shot Object Detection

Few-shot object detection has made substantial progressby representing novel class objects using the feature representation learned upon a set of base class objects. However,an implicit contradiction between novel class classification and representation is unfortunately ignored. On the one hand, to achieve accurate novel class classification, the distributions of either two base classes must be far away fromeach other (max-margin). On the other hand, to precisely represent novel classes, the distributions of base classes should be close to each other to reduce the intra-class distance of novel classes (min-margin). In this paper, we propose a class margin equilibrium (CME) approach, with the aim to optimize both feature space partition and novel class reconstruction in a systematic way. CME first converts the few-shot detection problem to the few-shot classification problem by using a fully connected layer to decouple localization features. CME then reserves adequate margin space for novel classes by introducing simple-yet-effective class margin loss during feature learning. Finally, CME pursues margin equilibrium by disturbing the features of novel class instances in an adversarial min-max fashion. Experiments on Pascal VOC and MS-COCO datasets show that CME significantly improves upon two baseline detectors (up to $3\sim 5\%$ in average), achieving state-of-the-art performance. Code is available at https://github.com/Bohao-Lee/CME .

cs.CV↗

arXiv2020

Prototype Mixture Models for Few-shot Semantic Segmentation

Few-shot segmentation is challenging because objects within the support and query images could significantly differ in appearance and pose. Using a single prototype acquired directly from the support image to segment the query image causes semantic ambiguity. In this paper, we propose prototype mixture models (PMMs), which correlate diverse image regions with multiple prototypes to enforce the prototype-based semantic representation. Estimated by an Expectation-Maximization algorithm, PMMs incorporate rich channel-wised and spatial semantics from limited support images. Utilized as representations as well as classifiers, PMMs fully leverage the semantics to activate objects in the query image while depressing background regions in a duplex manner. Extensive experiments on Pascal VOC and MS-COCO datasets show that PMMs significantly improve upon state-of-the-arts. Particularly, PMMs improve 5-shot segmentation performance on MS-COCO by up to 5.82\% with only a moderate cost for model size and inference speed.

cs.CV↗

arXiv2019

Method of diagnosing heart disease based on deep learning ECG signal

The traditional method of diagnosing heart disease on ECG signal is artificial observation. Some have tried to combine expertise and signal processing to classify ECG signal by heart disease type. However, the currency is not so sufficient that it can be used in medical applications. We develop an algorithm that combines signal processing and deep learning to classify ECG signals into Normal AF other rhythm and noise, which help us solve this problem. It is demonstrated that we can obtain the time-frequency diagram of ECG signal by wavelet transform, and use DNN to classify the time-frequency diagram to find out the heart disease that the signal collector may have. Overall, an accuracy of 94 percent is achieved on the validation set. According to the evaluation criteria of PhysioNet/Computing in Cardiology (CinC) in 2017, the F1 score of this method is 0.957, which is higher than the first place in the competition in 2017.

eess.SP↗