Mingbo Zhao

Jiaer Xia, Lei Tan, Pingyang Dai et al. · mila

Occluded person re-identification (Re-ID) aims to address the potential occlusion problem when matching occluded or holistic pedestrians from different camera views. Many methods use the background as artificial occlusion and rely on attention networks to exclude noisy interference. However, the significant discrepancy between simple background occlusion and realistic occlusion can negatively impact the generalization of the network. To address this issue, we propose a novel transformer-based Attention Disturbance and Dual-Path Constraint Network (ADP) to enhance the generalization of attention networks. Firstly, to imitate real-world obstacles, we introduce an Attention Disturbance Mask (ADM) module that generates an offensive noise, which can distract attention like a realistic occluder, as a more complex form of occlusion. Secondly, to fully exploit these complex occluded images, we develop a Dual-Path Constraint Module (DPC) that can obtain preferable supervision information from holistic images through dual-path interaction. With our proposed method, the network can effectively circumvent a wide variety of occlusions using the basic ViT baseline. Comprehensive experimental evaluations conducted on person re-ID benchmarks demonstrate the superiority of ADP over state-of-the-art methods.

Bo Wang, Zhao Zhang, Mingbo Zhao et al.

Mainstream image caption models are usually two-stage captioners, i.e., calculating object features by pre-trained detector, and feeding them into a language model to generate text descriptions. However, such an operation will cause a task-based information gap to decrease the performance, since the object features in detection task are suboptimal representation and cannot provide all necessary information for subsequent text generation. Besides, object features are usually represented by the last layer features that lose the local details of input images. In this paper, we propose a novel One-Stage Image Captioner (OSIC) with dynamic multi-sight learning, which directly transforms input image into descriptive sentences in one stage. As a result, the task-based information gap can be greatly reduced. To obtain rich features, we use the Swin Transformer to calculate multi-level features, and then feed them into a novel dynamic multi-sight embedding module to exploit both global structure and local texture of input images. To enhance the global modeling of encoder for caption, we propose a new dual-dimensional refining module to non-locally model the interaction of the embedded features. Finally, OSIC can obtain rich and useful information to improve the image caption task. Extensive comparisons on benchmark MS-COCO dataset verified the superior performance of our method.

Jiale Song, Jiaxin Luo, Xue-song Tang et al.

Large Vision-Language Models (LVLMs) achieve strong performance on many multimodal tasks, but object hallucinations severely undermine their reliability. Most existing studies focus on the text modality, attributing hallucinations to overly strong language priors and insufficient visual grounding. In contrast, we observe that abnormal attention patterns within the visual modality can also give rise to hallucinated objects. Building on this observation, we propose Segmentation-based Attention Entropy (SAE), which leverages semantic segmentation to quantify visual attention uncertainty in an object-level semantic space. Based on SAE, we further design a reliability score for hallucination detection and an SAE-guided attention adjustment method that modifies visual attention at inference time to mitigate hallucinations. We evaluate our approach on public benchmarks and in real embodied multimodal scenarios with quadruped robots. Experimental results show that SAE substantially reduces object hallucinations without any additional training cost, thereby enabling more trustworthy LVLM-driven perception and decision-making.

Yuan Gao, Jiayi Ma, Mingbo Zhao et al.

In this paper, we propose a novel Convolutional Neural Network (CNN) structure for general-purpose multi-task learning (MTL), which enables automatic feature fusing at every layer from different tasks. This is in contrast with the most widely used MTL CNN structures which empirically or heuristically share features on some specific layers (e.g., share all the features except the last convolutional layer). The proposed layerwise feature fusing scheme is formulated by combining existing CNN components in a novel way, with clear mathematical interpretability as discriminative dimensionality reduction, which is referred to as Neural Discriminative Dimensionality Reduction (NDDR). Specifically, we first concatenate features with the same spatial resolution from different tasks according to their channel dimension. Then, we show that the discriminative dimensionality reduction can be fulfilled by 1x1 Convolution, Batch Normalization, and Weight Decay in one CNN. The use of existing CNN components ensures the end-to-end training and the extensibility of the proposed NDDR layer to various state-of-the-art CNN architectures in a "plug-and-play" manner. The detailed ablation analysis shows that the proposed NDDR layer is easy to train and also robust to different hyperparameters. Experiments on different task sets with various base network architectures demonstrate the promising performance and desirable generalizability of our proposed method. The code of our paper is available at https://github.com/ethanygao/NDDR-CNN.

Jiacui Huang, Hongtao Zhang, Mingbo Zhao et al.

Vision-and-Language Navigation (VLN) is a challenging task that requires a robot to navigate in photo-realistic environments with human natural language promptings. Recent studies aim to handle this task by constructing the semantic spatial map representation of the environment, and then leveraging the strong ability of reasoning in large language models for generalizing code for guiding the robot navigation. However, these methods face limitations in instance-level and attribute-level navigation tasks as they cannot distinguish different instances of the same object. To address this challenge, we propose a new method, namely, Instance-aware Visual Language Map (IVLMap), to empower the robot with instance-level and attribute-level semantic mapping, where it is autonomously constructed by fusing the RGBD video data collected from the robot agent with special-designed natural language map indexing in the bird's-in-eye view. Such indexing is instance-level and attribute-level. In particular, when integrated with a large language model, IVLMap demonstrates the capability to i) transform natural language into navigation targets with instance and attribute information, enabling precise localization, and ii) accomplish zero-shot end-to-end navigation tasks based on natural language commands. Extensive navigation experiments are conducted. Simulation results illustrate that our method can achieve an average improvement of 14.4\% in navigation accuracy. Code and demo are released at https://ivlmap.github.io/.

Zhihao Tang, Shenghao Yang, Hongtao Zhang et al.

Traditional 3D garment creation requires extensive manual operations, resulting in time and labor costs. Recently, 3D Gaussian Splatting has achieved breakthrough progress in 3D scene reconstruction and rendering, attracting widespread attention and opening new pathways for 3D garment reconstruction. However, due to the unstructured and irregular nature of Gaussian primitives, it is difficult to reconstruct high-fidelity, non-watertight 3D garments. In this paper, we present GarmentGS, a dense point cloud-guided method that can reconstruct high-fidelity garment surfaces with high geometric accuracy and generate non-watertight, single-layer meshes. Our method introduces a fast dense point cloud reconstruction module that can complete garment point cloud reconstruction in 10 minutes, compared to traditional methods that require several hours. Furthermore, we use dense point clouds to guide the movement, flattening, and rotation of Gaussian primitives, enabling better distribution on the garment surface to achieve superior rendering effects and geometric accuracy. Through numerical and visual comparisons, our method achieves fast training and real-time rendering while maintaining competitive quality.

Lu Cheng, Mingbo Zhao

Human video instance segmentation plays an important role in computer understanding of human activities and is widely used in video processing, video surveillance, and human modeling in virtual reality. Most current VIS methods are based on Mask-RCNN framework, where the target appearance and motion information for data matching will increase computational cost and have an impact on segmentation real-time performance; on the other hand, the existing datasets for VIS focus less on all the people appearing in the video. In this paper, to solve the problems, we develop a new method for human video instance segmentation based on single-stage detector. To tracking the instance across the video, we have adopted data association strategy for matching the same instance in the video sequence, where we jointly learn target instance appearances and their affinities in a pair of video frames in an end-to-end fashion. We have also adopted the centroid sampling strategy for enhancing the embedding extraction ability of instance, which is to bias the instance position to the inside of each instance mask with heavy overlap condition. As a result, even there exists a sudden change in the character activity, the instance position will not move out of the mask, so that the problem that the same instance is represented by two different instances can be alleviated. Finally, we collect PVIS dataset by assembling several video instance segmentation datasets to fill the gap of the current lack of datasets dedicated to human video segmentation. Extensive simulations based on such dataset has been conduct. Simulation results verify the effectiveness and efficiency of the proposed work.

Yu Liu, Mingbo Zhao, Zhao Zhang et al.

Deep learning based virtual try-on system has achieved some encouraging progress recently, but there still remain several big challenges that need to be solved, such as trying on arbitrary clothes of all types, trying on the clothes from one category to another and generating image-realistic results with few artifacts. To handle this issue, we in this paper first collect a new dataset with all types of clothes, \ie tops, bottoms, and whole clothes, each one has multiple categories with rich information of clothing characteristics such as patterns, logos, and other details. Based on this dataset, we then propose the Arbitrary Virtual Try-On Network (AVTON) that is utilized for all-type clothes, which can synthesize realistic try-on images by preserving and trading off characteristics of the target clothes and the reference person. Our approach includes three modules: 1) Limbs Prediction Module, which is utilized for predicting the human body parts by preserving the characteristics of the reference person. This is especially good for handling cross-category try-on task (\eg long sleeves \(\leftrightarrow\) short sleeves or long pants \(\leftrightarrow\) skirts, \etc), where the exposed arms or legs with the skin colors and details can be reasonably predicted; 2) Improved Geometric Matching Module, which is designed to warp clothes according to the geometry of the target person. We improve the TPS based warping method with a compactly supported radial function (Wendland's \(Ψ\)-function); 3) Trade-Off Fusion Module, which is to trade off the characteristics of the warped clothes and the reference person. This module is to make the generated try-on images look more natural and realistic based on a fine-tune symmetry of the network structure. Extensive simulations are conducted and our approach can achieve better performance compared with the state-of-the-art virtual try-on methods.

Wenlong Cheng, Mingbo Zhao, Zhiling Ye et al.

Generative adversarial networks (GANs) have promoted remarkable advances in single-image super-resolution (SR) by recovering photo-realistic images. However, high memory consumption of GAN-based SR (usually generators) causes performance degradation and more energy consumption, hindering the deployment of GAN-based SR into resource-constricted mobile devices. In this paper, we propose a novel compression framework \textbf{M}ulti-scale \textbf{F}eature \textbf{A}ggregation Net based \textbf{GAN} (MFAGAN) for reducing the memory access cost of the generator. First, to overcome the memory explosion of dense connections, we utilize a memory-efficient multi-scale feature aggregation net as the generator. Second, for faster and more stable training, our method introduces the PatchGAN discriminator. Third, to balance the student discriminator and the compressed generator, we distill both the generator and the discriminator. Finally, we perform a hardware-aware neural architecture search (NAS) to find a specialized SubGenerator for the target mobile phone. Benefiting from these improvements, the proposed MFAGAN achieves up to \textbf{8.3}$\times$ memory saving and \textbf{42.9}$\times$ computation reduction, with only minor visual quality degradation, compared with ESRGAN. Empirical studies also show $\sim$\textbf{70} milliseconds latency on Qualcomm Snapdragon 865 chipset.

Jicong Fan, Yiheng Tu, Zhao Zhang et al.

The performance of spectral clustering heavily relies on the quality of affinity matrix. A variety of affinity-matrix-construction (AMC) methods have been proposed but they have hyperparameters to determine beforehand, which requires strong experience and leads to difficulty in real applications, especially when the inter-cluster similarity is high and/or the dataset is large. In addition, we often need to choose different AMC methods for different datasets, which still depends on experience. To solve these two challenging problems, in this paper, we present a simple yet effective method for automated spectral clustering. First, we propose to find the most reliable affinity matrix via grid search or Bayesian optimization among a set of candidates given by different AMC methods with different hyperparameters, where the reliability is quantified by the \textit{relative-eigen-gap} of graph Laplacian introduced in this paper. Second, we propose a fast and accurate AMC method based on least squares representation and thresholding and prove its effectiveness theoretically. Finally, we provide a large-scale extension for the automated spectral clustering method, of which the time complexity is linear with the number of data points. Extensive experiments of natural image clustering show that our method is more versatile, accurate, and efficient than baseline methods.

Yanyan Wei, Zhao Zhang, Yang Wang et al.

Removing the rain streaks from single image is still a challenging task, since the shapes and directions of rain streaks in the synthetic datasets are very different from real images. Although supervised deep deraining networks have obtained impressive results on synthetic datasets, they still cannot obtain satisfactory results on real images due to weak generalization of rain removal capacity, i.e., the pre-trained models usually cannot handle new shapes and directions that may lead to over-derained/under-derained results. In this paper, we propose a new semi-supervised GAN-based deraining network termed Semi-DerainGAN, which can use both synthetic and real rainy images in a uniform network using two supervised and unsupervised processes. Specifically, a semi-supervised rain streak learner termed SSRML sharing the same parameters of both processes is derived, which makes the real images contribute more rain streak information. To deliver better deraining results, we design a paired discriminator for distinguishing the real pairs from fake pairs. Note that we also contribute a new real-world rainy image dataset Real200 to alleviate the difference between the synthetic and real image do-mains. Extensive results on public datasets show that our model can obtain competitive performance, especially on real images.

Yan Zhang, Zhao Zhang, Zheng Zhang et al.

In this paper, we investigate the unsupervised deep representation learning issue and technically propose a novel framework called Deep Self-representative Concept Factorization Network (DSCF-Net), for clustering deep features. To improve the representation and clustering abilities, DSCF-Net explicitly considers discovering hidden deep semantic features, enhancing the robustness proper-ties of the deep factorization to noise and preserving the local man-ifold structures of deep features. Specifically, DSCF-Net seamlessly integrates the robust deep concept factorization, deep self-expressive representation and adaptive locality preserving feature learning into a unified framework. To discover hidden deep repre-sentations, DSCF-Net designs a hierarchical factorization architec-ture using multiple layers of linear transformations, where the hierarchical representation is performed by formulating the prob-lem as optimizing the basis concepts in each layer to improve the representation indirectly. DSCF-Net also improves the robustness by subspace recovery for sparse error correction firstly and then performs the deep factorization in the recovered visual subspace. To obtain locality-preserving representations, we also present an adaptive deep self-representative weighting strategy by using the coefficient matrix as the adaptive reconstruction weights to keep the locality of representations. Extensive comparison results with several other related models show that DSCF-Net delivers state-of-the-art performance on several public databases.

Huan Zhang, Zhao Zhang, Mingbo Zhao et al.

The graph-based semi-supervised label propagation algorithm has delivered impressive classification results. However, the estimated soft labels typically contain mixed signs and noise, which cause inaccurate predictions due to the lack of suitable constraints. Moreover, available methods typically calculate the weights and estimate the labels in the original input space, which typically contains noise and corruption. Thus, the en-coded similarities and manifold smoothness may be inaccurate for label estimation. In this paper, we present effective schemes for resolving these issues and propose a novel and robust semi-supervised classification algorithm, namely, the tri-ple-matrix-recovery-based robust auto-weighted label propa-gation framework (ALP-TMR). Our ALP-TMR introduces a triple matrix recovery mechanism to remove noise or mixed signs from the estimated soft labels and improve the robustness to noise and outliers in the steps of assigning weights and pre-dicting the labels simultaneously. Our method can jointly re-cover the underlying clean data, clean labels and clean weighting spaces by decomposing the original data, predicted soft labels or weights into a clean part plus an error part by fitting noise. In addition, ALP-TMR integrates the au-to-weighting process by minimizing reconstruction errors over the recovered clean data and clean soft labels, which can en-code the weights more accurately to improve both data rep-resentation and classification. By classifying samples in the recovered clean label and weight spaces, one can potentially improve the label prediction results. The results of extensive experiments demonstrated the satisfactory performance of our ALP-TMR.

Zhao Zhang, Lei Jia, Mingbo Zhao et al.

Kernel methods have been successfully applied to the areas of pattern recognition and data mining. In this paper, we mainly discuss the issue of propagating labels in kernel space. A Kernel-Induced Label Propagation (Kernel-LP) framework by mapping is proposed for high-dimensional data classification using the most informative patterns of data in kernel space. The essence of Kernel-LP is to perform joint label propagation and adaptive weight learning in a transformed kernel space. That is, our Kernel-LP changes the task of label propagation from the commonly-used Euclidean space in most existing work to kernel space. The motivation of our Kernel-LP to propagate labels and learn the adaptive weights jointly by the assumption of an inner product space of inputs, i.e., the original linearly inseparable inputs may be mapped to be separable in kernel space. Kernel-LP is based on existing positive and negative LP model, i.e., the effects of negative label information are integrated to improve the label prediction power. Also, Kernel-LP performs adaptive weight construction over the same kernel space, so it can avoid the tricky process of choosing the optimal neighborhood size suffered in traditional criteria. Two novel and efficient out-of-sample approaches for our Kernel-LP to involve new test data are also presented, i.e., (1) direct kernel mapping and (2) kernel mapping-induced label reconstruction, both of which purely depend on the kernel matrix between training set and testing set. Owing to the kernel trick, our algorithms will be applicable to handle the high-dimensional real data. Extensive results on real datasets demonstrate the effectiveness of our approach.