arXiv subjects

Jifeng Ning

Publications and source records attributed to Jifeng Ning.

11 recordsLinked to original sources

arXiv2026

Uncovering and Mitigating Positional Blind Spots in Vision-Language-Action Models

Recent Vision-Language-Action (VLA) models achieve promising performance in robotic manipulation, typically measured by success rates aggregated over predefined object configurations, an evaluation that implicitly assumes spatially uniform competence across the workspace. However, this assumption does not hold: even with the instruction and every other scene factor held fixed, merely relocating a task-irrelevant distractor can sharply raise the failure probability within localized, spatially coherent regions, which we term Positional Blind Spots (PBS). In this paper, we propose a two-stage black-box framework to uncover and mitigate PBS. During the uncovering stage, we grid the workspace and apply a one-sided log-likelihood-ratio test to localize PBS cells with significantly elevated risk. During the mitigation stage, we fine-tune the policy via LoRA on demonstrations collected from these PBS regions, improving competence there while largely preserving performance across the rest of the workspace. We evaluate our framework on five state-of-the-art VLA policies across two benchmarks, and find that PBS are pervasive and spatially concentrated in all of them, with failure rates up to 0.58. Our search strategy achieves an average F1-score of 0.678, outperforming random search and adaptive sampling baselines by 0.268 and 0.178, respectively. Guided by the discovered regions, targeted fine-tuning reduces the overall failure rate by 40.00%--85.19%.

cs.RO↗

arXiv2026

Exploring Modality-Aware Fusion and Decoupled Temporal Propagation for Multi-Modal Object Tracking

Most existing multimodal trackers adopt uniform fusion strategies, overlooking the inherent differences between modalities. Moreover, they propagate temporal information through mixed tokens, leading to entangled and less discriminative temporal representations. To address these limitations, we propose MDTrack, a novel framework for modality aware fusion and decoupled temporal propagation in multimodal object tracking. Specifically, for modality aware fusion, we allocate dedicated experts to each modality, including infrared, event, depth, and RGB, to process their respective representations. The gating mechanism within the Mixture of Experts dynamically selects the optimal experts based on the input features, enabling adaptive and modality specific fusion. For decoupled temporal propagation, we introduce two separate State Space Model structures to independently store and update the hidden states of the RGB and X modal streams, effectively capturing their distinct temporal information. To ensure synergy between the two temporal representations, we incorporate a set of cross attention modules between the input features of the two SSMs, facilitating implicit information exchange. The resulting temporally enriched features are then integrated into the backbone through another set of cross attention modules, enhancing MDTrack's ability to leverage temporal information. Extensive experiments demonstrate the effectiveness of our proposed method. Both MDTrack S and MDTrack U achieve state of the art performance across five multimodal tracking benchmarks.

cs.CV↗

arXiv2025

DIDiffGes: Decoupled Semi-Implicit Diffusion Models for Real-time Gesture Generation from Speech

Diffusion models have demonstrated remarkable synthesis quality and diversity in generating co-speech gestures. However, the computationally intensive sampling steps associated with diffusion models hinder their practicality in real-world applications. Hence, we present DIDiffGes, for a Decoupled Semi-Implicit Diffusion model-based framework, that can synthesize high-quality, expressive gestures from speech using only a few sampling steps. Our approach leverages Generative Adversarial Networks (GANs) to enable large-step sampling for diffusion model. We decouple gesture data into body and hands distributions and further decompose them into marginal and conditional distributions. GANs model the marginal distribution implicitly, while L2 reconstruction loss learns the conditional distributions exciplictly. This strategy enhances GAN training stability and ensures expressiveness of generated full-body gestures. Our framework also learns to denoise root noise conditioned on local body representation, guaranteeing stability and realism. DIDiffGes can generate gestures from speech with just 10 sampling steps, without compromising quality and expressiveness, reducing the number of sampling steps by a factor of 100 compared to existing methods. Our user study reveals that our method outperforms state-of-the-art approaches in human likeness, appropriateness, and style correctness. Project is https://cyk990422.github.io/DIDiffGes.

cs.GR↗

arXiv2024

Conditional GAN for Enhancing Diffusion Models in Efficient and Authentic Global Gesture Generation from Audios

Audio-driven simultaneous gesture generation is vital for human-computer communication, AI games, and film production. While previous research has shown promise, there are still limitations. Methods based on VAEs are accompanied by issues of local jitter and global instability, whereas methods based on diffusion models are hampered by low generation efficiency. This is because the denoising process of DDPM in the latter relies on the assumption that the noise added at each step is sampled from a unimodal distribution, and the noise values are small. DDIM borrows the idea from the Euler method for solving differential equations, disrupts the Markov chain process, and increases the noise step size to reduce the number of denoising steps, thereby accelerating generation. However, simply increasing the step size during the step-by-step denoising process causes the results to gradually deviate from the original data distribution, leading to a significant drop in the quality of the generated actions and the emergence of unnatural artifacts. In this paper, we break the assumptions of DDPM and achieves breakthrough progress in denoising speed and fidelity. Specifically, we introduce a conditional GAN to capture audio control signals and implicitly match the multimodal denoising distribution between the diffusion and denoising steps within the same sampling step, aiming to sample larger noise values and apply fewer denoising steps for high-speed generation.

cs.SD↗

arXiv2024

ExpGest: Expressive Speaker Generation Using Diffusion Model and Hybrid Audio-Text Guidance

Existing gesture generation methods primarily focus on upper body gestures based on audio features, neglecting speech content, emotion, and locomotion. These limitations result in stiff, mechanical gestures that fail to convey the true meaning of audio content. We introduce ExpGest, a novel framework leveraging synchronized text and audio information to generate expressive full-body gestures. Unlike AdaIN or one-hot encoding methods, we design a noise emotion classifier for optimizing adversarial direction noise, avoiding melody distortion and guiding results towards specified emotions. Moreover, aligning semantic and gestures in the latent space provides better generalization capabilities. ExpGest, a diffusion model-based gesture generation framework, is the first attempt to offer mixed generation modes, including audio-driven gestures and text-shaped motion. Experiments show that our framework effectively learns from combined text-driven motion and audio-induced gesture datasets, and preliminary results demonstrate that ExpGest achieves more expressive, natural, and controllable global motion in speakers compared to state-of-the-art models.

cs.SD↗

arXiv2023

Human-to-Human Interaction Detection

A comprehensive understanding of interested human-to-human interactions in video streams, such as queuing, handshaking, fighting and chasing, is of immense importance to the surveillance of public security in regions like campuses, squares and parks. Different from conventional human interaction recognition, which uses choreographed videos as inputs, neglects concurrent interactive groups, and performs detection and recognition in separate stages, we introduce a new task named human-to-human interaction detection (HID). HID devotes to detecting subjects, recognizing person-wise actions, and grouping people according to their interactive relations, in one model. First, based on the popular AVA dataset created for action detection, we establish a new HID benchmark, termed AVA-Interaction (AVA-I), by adding annotations on interactive relations in a frame-by-frame manner. AVA-I consists of 85,254 frames and 86,338 interactive groups, and each image includes up to 4 concurrent interactive groups. Second, we present a novel baseline approach SaMFormer for HID, containing a visual feature extractor, a split stage which leverages a Transformer-based model to decode action instances and interactive groups, and a merging stage which reconstructs the relationship between instances and groups. All SaMFormer components are jointly trained in an end-to-end manner. Extensive experiments on AVA-I validate the superiority of SaMFormer over representative methods. The dataset and code will be made public to encourage more follow-up studies.

cs.CV↗

arXiv2023

ReBaR: Reference-Based Reasoning for Robust Pose Estimation from Monocular Images

R}easoning for Robust Human Pose and Shape Estimation), designed to estimate human body shape and pose from single-view images. ReBaR effectively addresses the challenges of occlusions and depth ambiguity by learning reference features for part regression reasoning. Our approach starts by extracting features from both body and part regions using an attention-guided mechanism. Subsequently, these features are used to encode additional part-body dependencies for individual part regression, with part features serving as queries and the body feature as a reference. This reference-based reasoning allows our network to infer the spatial relationships of occluded parts with the body, utilizing visible parts and body reference information. ReBaR outperforms contemporary methods on three benchmark datasets and still maintains competitive advantages among recent new approaches. Demonstrating significant improvement in handling depth ambiguity and occlusion. These results strongly support the effectiveness of our reference-based framework for estimating human body shape and pose from single-view images.

cs.CV↗

arXiv2020

Enhanced nonconvex low-rank approximation of tensor multi-modes for tensor completion

Higher-order low-rank tensor arises in many data processing applications and has attracted great interests. Inspired by low-rank approximation theory, researchers have proposed a series of effective tensor completion methods. However, most of these methods directly consider the global low-rankness of underlying tensors, which is not sufficient for a low sampling rate; in addition, the single nuclear norm or its relaxation is usually adopted to approximate the rank function, which would lead to suboptimal solution deviated from the original one. To alleviate the above problems, in this paper, we propose a novel low-rank approximation of tensor multi-modes (LRATM), in which a double nonconvex $L_γ$ norm is designed to represent the underlying joint-manifold drawn from the modal factorization factors of the underlying tensor. A block successive upper-bound minimization method-based algorithm is designed to efficiently solve the proposed model, and it can be demonstrated that our numerical scheme converges to the coordinatewise minimizers. Numerical results on three types of public multi-dimensional datasets have tested and shown that our algorithm can recover a variety of low-rank tensors with significantly fewer samples than the compared methods.

cs.CV↗

arXiv2020

Hyperspectral Image Denoising via Global Spatial-Spectral Total Variation Regularized Nonconvex Local Low-Rank Tensor Approximation

Hyperspectral image (HSI) denoising aims to restore clean HSI from the noise-contaminated one. Noise contamination can often be caused during data acquisition and conversion. In this paper, we propose a novel spatial-spectral total variation (SSTV) regularized nonconvex local low-rank (LR) tensor approximation method to remove mixed noise in HSIs. From one aspect, the clean HSI data have its underlying local LR tensor property, even though the real HSI data may not be globally low-rank due to out-liers and non-Gaussian noise. According to this fact, we propose a novel tensor $L_γ$-norm to formulate the local LR prior. From another aspect, HSIs are assumed to be piecewisely smooth in the global spatial and spectral domains. Instead of traditional bandwise total variation, we use the SSTV regularization to simultaneously consider global spatial structure and spectral correlation of neighboring bands. Results on simulated and real HSI datasets indicate that the use of local LR tensor penalty and global SSTV can boost the preserving of local details and overall structural information in HSIs.

eess.IV↗

arXiv2020

Hyperspectral Image Restoration via Global Total Variation Regularized Local nonconvex Low-Rank matrix Approximation

Several bandwise total variation (TV) regularized low-rank (LR)-based models have been proposed to remove mixed noise in hyperspectral images (HSIs). Conventionally, the rank of LR matrix is approximated using nuclear norm (NN). The NN is defined by adding all singular values together, which is essentially a $L_1$-norm of the singular values. It results in non-negligible approximation errors and thus the resulting matrix estimator can be significantly biased. Moreover, these bandwise TV-based methods exploit the spatial information in a separate manner. To cope with these problems, we propose a spatial-spectral TV (SSTV) regularized non-convex local LR matrix approximation (NonLLRTV) method to remove mixed noise in HSIs. From one aspect, local LR of HSIs is formulated using a non-convex $L_γ$-norm, which provides a closer approximation to the matrix rank than the traditional NN. From another aspect, HSIs are assumed to be piecewisely smooth in the global spatial domain. The TV regularization is effective in preserving the smoothness and removing Gaussian noise. These facts inspire the integration of the NonLLR with TV regularization. To address the limitations of bandwise TV, we use the SSTV regularization to simultaneously consider global spatial structure and spectral correlation of neighboring bands. Experiment results indicate that the use of local non-convex penalty and global SSTV can boost the preserving of spatial piecewise smoothness and overall structural information.

cs.CV↗

arXiv2020

Tensor completion using enhanced multiple modes low-rank prior and total variation

In this paper, we propose a novel model to recover a low-rank tensor by simultaneously performing double nuclear norm regularized low-rank matrix factorizations to the all-mode matricizations of the underlying tensor. An block successive upper-bound minimization algorithm is applied to solve the model. Subsequence convergence of our algorithm can be established, and our algorithm converges to the coordinate-wise minimizers in some mild conditions. Several experiments on three types of public data sets show that our algorithm can recover a variety of low-rank tensors from significantly fewer samples than the other testing tensor completion methods.

cs.CV↗