arXiv subjects

Xuming He

Publications and source records attributed to Xuming He.

At least 73 records · Page 4Linked to original sources

arXiv2023

Gradient-Map-Guided Adaptive Domain Generalization for Cross Modality MRI Segmentation

Cross-modal MRI segmentation is of great value for computer-aided medical diagnosis, enabling flexible data acquisition and model generalization. However, most existing methods have difficulty in handling local variations in domain shift and typically require a significant amount of data for training, which hinders their usage in practice. To address these problems, we propose a novel adaptive domain generalization framework, which integrates a learning-free cross-domain representation based on image gradient maps and a class prior-informed test-time adaptation strategy for mitigating local domain shift. We validate our approach on two multi-modal MRI datasets with six cross-modal segmentation tasks. Across all the task settings, our method consistently outperforms competing approaches and shows a stable performance even with limited training data.

cs.CV↗

arXiv2023

SPHINX: The Joint Mixing of Weights, Tasks, and Visual Embeddings for Multi-modal Large Language Models

We present SPHINX, a versatile multi-modal large language model (MLLM) with a joint mixing of model weights, tuning tasks, and visual embeddings. First, for stronger vision-language alignment, we unfreeze the large language model (LLM) during pre-training, and introduce a weight mix strategy between LLMs trained by real-world and synthetic data. By directly integrating the weights from two domains, the mixed LLM can efficiently incorporate diverse semantics with favorable robustness. Then, to enable multi-purpose capabilities, we mix a variety of tasks for joint visual instruction tuning, and design task-specific instructions to avoid inter-task conflict. In addition to the basic visual question answering, we include more challenging tasks such as region-level understanding, caption grounding, document layout detection, and human pose estimation, contributing to mutual enhancement over different scenarios. Additionally, we propose to extract comprehensive visual embeddings from various network architectures, pre-training paradigms, and information granularity, providing language models with more robust image representations. Based on our proposed joint mixing, SPHINX exhibits superior multi-modal understanding capabilities on a wide range of applications. On top of this, we further propose an efficient strategy aiming to better capture fine-grained appearances of high-resolution images. With a mixing of different scales and high-resolution sub-images, SPHINX attains exceptional visual parsing and reasoning performance on existing evaluation benchmarks. We hope our work may cast a light on the exploration of joint mixing in future MLLM research. Code is released at https://github.com/Alpha-VLLM/LLaMA2-Accessory.

cs.CV↗

arXiv2023

Optical ReLU-like Activation Function Based on a Semiconductor Laser with Optical Injection

Artificial neural networks usually consist of successive linear multiply-accumulate operations and nonlinear activation functions. However, most optical neural networks only achieve the linear operation in the optical domain, while the optical implementation of activation function remains challenging. Here we present an optical ReLU-like activation function based on a semiconductor laser subject to the optical injection in experiment. The ReLU-like function is achieved in a broad regime above the Hopf bifurcation of the injection-locking diagram. In particular, the slope of the activation function is reconfigurable by tuning the frequency difference between the master laser and the slave laser.

physics.optics↗

arXiv2023

Cascaded Sparse Feature Propagation Network for Interactive Segmentation

We aim to tackle the problem of point-based interactive segmentation, in which the key challenge is to propagate the user-provided annotations to unlabeled regions efficiently. Existing methods tackle this challenge by utilizing computationally expensive fully connected graphs or transformer architectures that sacrifice important fine-grained information required for accurate segmentation. To overcome these limitations, we propose a cascade sparse feature propagation network that learns a click-augmented feature representation for propagating user-provided information to unlabeled regions. The sparse design of our network enables efficient information propagation on high-resolution features, resulting in more detailed object segmentation. We validate the effectiveness of our method through comprehensive experiments on various benchmarks, and the results demonstrate the superior performance of our approach. Code is available at \href{https://github.com/kleinzcy/CSFPN}{https://github.com/kleinzcy/CSFPN}.

cs.CV↗

arXiv2023

ATTA: Anomaly-aware Test-Time Adaptation for Out-of-Distribution Detection in Segmentation

Recent advancements in dense out-of-distribution (OOD) detection have primarily focused on scenarios where the training and testing datasets share a similar domain, with the assumption that no domain shift exists between them. However, in real-world situations, domain shift often exits and significantly affects the accuracy of existing out-of-distribution (OOD) detection models. In this work, we propose a dual-level OOD detection framework to handle domain shift and semantic shift jointly. The first level distinguishes whether domain shift exists in the image by leveraging global low-level features, while the second level identifies pixels with semantic shift by utilizing dense high-level feature maps. In this way, we can selectively adapt the model to unseen domains as well as enhance model's capacity in detecting novel classes. We validate the efficacy of our proposed method on several OOD segmentation benchmarks, including those with significant domain shifts and those without, observing consistent performance improvements across various baseline models. Code is available at ${\href{https://github.com/gaozhitong/ATTA}{https://github.com/gaozhitong/ATTA}}$.

cs.CV↗

arXiv2023

The Robust Semantic Segmentation UNCV2023 Challenge Results

This paper outlines the winning solutions employed in addressing the MUAD uncertainty quantification challenge held at ICCV 2023. The challenge was centered around semantic segmentation in urban environments, with a particular focus on natural adversarial scenarios. The report presents the results of 19 submitted entries, with numerous techniques drawing inspiration from cutting-edge uncertainty quantification methodologies presented at prominent conferences in the fields of computer vision and machine learning and journals over the past few years. Within this document, the challenge is introduced, shedding light on its purpose and objectives, which primarily revolved around enhancing the robustness of semantic segmentation in urban scenes under varying natural adversarial conditions. The report then delves into the top-performing solutions. Moreover, the document aims to provide a comprehensive overview of the diverse solutions deployed by all participants. By doing so, it seeks to offer readers a deeper insight into the array of strategies that can be leveraged to effectively handle the inherent uncertainties associated with autonomous driving and semantic segmentation, especially within urban environments.

cs.CV↗

arXiv2023

Class-relation Knowledge Distillation for Novel Class Discovery

We tackle the problem of novel class discovery, which aims to learn novel classes without supervision based on labeled data from known classes. A key challenge lies in transferring the knowledge in the known-class data to the learning of novel classes. Previous methods mainly focus on building a shared representation space for knowledge transfer and often ignore modeling class relations. To address this, we introduce a class relation representation for the novel classes based on the predicted class distribution of a model trained on known classes. Empirically, we find that such class relation becomes less informative during typical discovery training. To prevent such information loss, we propose a novel knowledge distillation framework, which utilizes our class-relation representation to regularize the learning of novel classes. In addition, to enable a flexible knowledge distillation scheme for each data point in novel classes, we develop a learnable weighting function for the regularization, which adaptively promotes knowledge transfer based on the semantic similarity between the novel and known classes. To validate the effectiveness and generalization of our method, we conduct extensive experiments on multiple benchmarks, including CIFAR100, Stanford Cars, CUB, and FGVC-Aircraft datasets. Our results demonstrate that the proposed method outperforms the previous state-of-the-art methods by a significant margin on almost all benchmarks. Code is available at \href{https://github.com/kleinzcy/Cr-KD-NCD}{here}.

cs.CV↗

arXiv2023

Novel Class Discovery for Long-tailed Recognition

While the novel class discovery has recently made great progress, existing methods typically focus on improving algorithms on class-balanced benchmarks. However, in real-world recognition tasks, the class distributions of their corresponding datasets are often imbalanced, which leads to serious performance degeneration of those methods. In this paper, we consider a more realistic setting for novel class discovery where the distributions of novel and known classes are long-tailed. One main challenge of this new problem is to discover imbalanced novel classes with the help of long-tailed known classes. To tackle this problem, we propose an adaptive self-labeling strategy based on an equiangular prototype representation of classes. Our method infers high-quality pseudo-labels for the novel classes by solving a relaxed optimal transport problem and effectively mitigates the class biases in learning the known and novel classes. We perform extensive experiments on CIFAR100, ImageNet100, Herbarium19 and large-scale iNaturalist18 datasets, and the results demonstrate the superiority of our method. Our code is available at https://github.com/kleinzcy/NCDLR.

cs.CV↗

arXiv2023

Grounded Image Text Matching with Mismatched Relation Reasoning

This paper introduces Grounded Image Text Matching with Mismatched Relation (GITM-MR), a novel visual-linguistic joint task that evaluates the relation understanding capabilities of transformer-based pre-trained models. GITM-MR requires a model to first determine if an expression describes an image, then localize referred objects or ground the mismatched parts of the text. We provide a benchmark for evaluating pre-trained models on this task, with a focus on the challenging settings of limited data and out-of-distribution sentence lengths. Our evaluation demonstrates that pre-trained models lack data efficiency and length generalization ability. To address this, we propose the Relation-sensitive Correspondence Reasoning Network (RCRN), which incorporates relation-aware reasoning via bi-directional message propagation guided by language structure. RCRN can be interpreted as a modular program and delivers strong performance in both length generalization and data efficiency.

cs.CV↗

arXiv2023

HOICLIP: Efficient Knowledge Transfer for HOI Detection with Vision-Language Models

Human-Object Interaction (HOI) detection aims to localize human-object pairs and recognize their interactions. Recently, Contrastive Language-Image Pre-training (CLIP) has shown great potential in providing interaction prior for HOI detectors via knowledge distillation. However, such approaches often rely on large-scale training data and suffer from inferior performance under few/zero-shot scenarios. In this paper, we propose a novel HOI detection framework that efficiently extracts prior knowledge from CLIP and achieves better generalization. In detail, we first introduce a novel interaction decoder to extract informative regions in the visual feature map of CLIP via a cross-attention mechanism, which is then fused with the detection backbone by a knowledge integration block for more accurate human-object pair detection. In addition, prior knowledge in CLIP text encoder is leveraged to generate a classifier by embedding HOI descriptions. To distinguish fine-grained interactions, we build a verb classifier from training data via visual semantic arithmetic and a lightweight verb representation adapter. Furthermore, we propose a training-free enhancement to exploit global HOI predictions from CLIP. Extensive experiments demonstrate that our method outperforms the state of the art by a large margin on various settings, e.g. +4.04 mAP on HICO-Det. The source code is available in https://github.com/Artanic30/HOICLIP.

cs.CV↗

arXiv2023

Human-centric Scene Understanding for 3D Large-scale Scenarios

Human-centric scene understanding is significant for real-world applications, but it is extremely challenging due to the existence of diverse human poses and actions, complex human-environment interactions, severe occlusions in crowds, etc. In this paper, we present a large-scale multi-modal dataset for human-centric scene understanding, dubbed HuCenLife, which is collected in diverse daily-life scenarios with rich and fine-grained annotations. Our HuCenLife can benefit many 3D perception tasks, such as segmentation, detection, action recognition, etc., and we also provide benchmarks for these tasks to facilitate related research. In addition, we design novel modules for LiDAR-based segmentation and action recognition, which are more applicable for large-scale human-centric scenarios and achieve state-of-the-art performance.

cs.CV↗

arXiv2023

A Physics-Informed Data-Driven Fault Location Method for Transmission Lines Using Single-Ended Measurements with Field Data Validation

Data driven transmission line fault location methods have the potential to more accurately locate faults by extracting fault information from available data. However, most of the data driven fault location methods in the literature are not validated by field data for the following reasons. On one hand, the available field data during faults are very limited for one specific transmission line, and using field data for training is close to impossible. On the other hand, if simulation data are utilized for training, the mismatch between the simulation system and the practical system will cause fault location errors. To this end, this paper proposes a physics-informed data-driven fault location method. The data from a practical fault event are first analyzed to extract the ranges of system and fault parameters such as equivalent source impedances, loading conditions, fault inception angles (FIA) and fault resistances. Afterwards, the simulation system is constructed with the ranges of parameters, to generate data for training. This procedure merges the gap between simulation and practical power systems, and at the same time considers the uncertainty of system and fault parameters in practice. The proposed data-driven method does not require system parameters, only requires instantaneous voltage and current measurements at the local terminal, with a low sampling rate of several kHz and a short fault time window of half a cycle before and after the fault occurs. Numerical experiments and field data experiments clearly validate the advantages of the proposed method over existing data driven methods.

eess.SY↗

arXiv2023

Scalable wavelength-multiplexing photonic reservoir computing

Photonic reservoir computing (PRC) is a special hardware recurrent neural network, which is featured with fast training speed and low training cost. This work shows a wavelength-multiplexing PRC architecture, taking advantage of the numerous longitudinal modes in a Fabry-Perot semiconductor laser. These modes construct connected physical neurons in parallel, while an optical feedback loop provides interactive virtual neurons in series. We experimentally demonstrate a four-channel wavelength-multiplexing PRC, which runs four times faster than the single-channel case. It is proved that the multiplexing PRC exhibits superior performance on the task of signal equalization in an optical fiber communication link. Particularly, this scheme is highly scalable owing to the rich mode resources in Fabry-Perot lasers.

physics.optics↗

arXiv2023

Treatment Effect Estimation with Efficient Data Aggregation

Data aggregation, also known as meta analysis, is widely used to combine knowledge on parameters shared in common (e.g., average treatment effect) between multiple studies. In this paper, we introduce an attractive data aggregation scheme that pools summary statistics from various existing studies. Our scheme informs the design of new validation studies and yields us unbiased estimators for the shared parameters. In our setup, each existing study applies a LASSO regression to select a parsimonious model from a large set of covariates. It is well known that post-hoc estimators, in the selected model, tend to be biased. We show that a novel technique called \textit{data carving} yields us a new unbiased estimator by aggregating simple summary statistics from all existing studies. Our estimator has two key features: (a) we make the fullest possible use of data, from all studies, without the risk of bias from model selection; (b) we enjoy the added benefit of individual data privacy, because raw data from these studies need not be shared or stored for efficient estimation.

stat.ME↗

arXiv2023

Estimation of Complier Expected Shortfall Treatment Effects with a Binary Instrumental Variable

Estimating the causal effect of a treatment or exposure for a subpopulation is of great interest in many biomedical and economical studies. Expected shortfall, also referred to as the super-quantile, is an attractive effect-size measure that can accommodate data heterogeneity and aggregate local information of effect over a certain region of interest of the outcome distribution. In this article, we propose the ComplieR Expected Shortfall Treatment Effect (CRESTE) model under an instrumental variable framework to quantity the CRESTE for a binary endogenous treatment variable. By utilizing the special characteristics of a binary instrumental variable and a specific formulation of Neyman-orthogonalization, we propose a two-step estimation procedure, which can be implemented by simply solving weighted least-squares regression and weighted quantile regression with estimated weights. We develop the asymptotic properties for the proposed estimator and use numerical simulations to confirm its validity and robust finite-sample performance. An illustrative analysis of a National Job Training Partnership Act study is presented to show the practical utility of the proposed method.

stat.ME↗

arXiv2023

Weakly-supervised HOI Detection via Prior-guided Bi-level Representation Learning

Human object interaction (HOI) detection plays a crucial role in human-centric scene understanding and serves as a fundamental building-block for many vision tasks. One generalizable and scalable strategy for HOI detection is to use weak supervision, learning from image-level annotations only. This is inherently challenging due to ambiguous human-object associations, large search space of detecting HOIs and highly noisy training signal. A promising strategy to address those challenges is to exploit knowledge from large-scale pretrained models (e.g., CLIP), but a direct knowledge distillation strategy~\citep{liao2022gen} does not perform well on the weakly-supervised setting. In contrast, we develop a CLIP-guided HOI representation capable of incorporating the prior knowledge at both image level and HOI instance level, and adopt a self-taught mechanism to prune incorrect human-object associations. Experimental results on HICO-DET and V-COCO show that our method outperforms the previous works by a sizable margin, showing the efficacy of our HOI representation.

cs.CV↗

arXiv2023

Modeling Multimodal Aleatoric Uncertainty in Segmentation with Mixture of Stochastic Experts

Equipping predicted segmentation with calibrated uncertainty is essential for safety-critical applications. In this work, we focus on capturing the data-inherent uncertainty (aka aleatoric uncertainty) in segmentation, typically when ambiguities exist in input images. Due to the high-dimensional output space and potential multiple modes in segmenting ambiguous images, it remains challenging to predict well-calibrated uncertainty for segmentation. To tackle this problem, we propose a novel mixture of stochastic experts (MoSE) model, where each expert network estimates a distinct mode of the aleatoric uncertainty and a gating network predicts the probabilities of an input image being segmented in those modes. This yields an efficient two-level uncertainty representation. To learn the model, we develop a Wasserstein-like loss that directly minimizes the distribution distance between the MoSE and ground truth annotations. The loss can easily integrate traditional segmentation quality measures and be efficiently optimized via constraint relaxation. We validate our method on the LIDC-IDRI dataset and a modified multimodal Cityscapes dataset. Results demonstrate that our method achieves the state-of-the-art or competitive performance on all metrics.

cs.CV↗

arXiv2023

Dynamic Grained Encoder for Vision Transformers

Transformers, the de-facto standard for language modeling, have been recently applied for vision tasks. This paper introduces sparse queries for vision transformers to exploit the intrinsic spatial redundancy of natural images and save computational costs. Specifically, we propose a Dynamic Grained Encoder for vision transformers, which can adaptively assign a suitable number of queries to each spatial region. Thus it achieves a fine-grained representation in discriminative regions while keeping high efficiency. Besides, the dynamic grained encoder is compatible with most vision transformer frameworks. Without bells and whistles, our encoder allows the state-of-the-art vision transformers to reduce computational complexity by 40%-60% while maintaining comparable performance on image classification. Extensive experiments on object detection and segmentation further demonstrate the generalizability of our approach. Code is available at https://github.com/StevenGrove/vtpack.

cs.CV↗