Hongyuan Zhu

Chuangguan Ye, Hongyuan Zhu, Bo Zhang et al. · deepmind

In recent years, research on few-shot learning (FSL) has been fast-growing in the 2D image domain due to the less requirement for labeled training data and greater generalization for novel classes. However, its application in 3D point cloud data is relatively under-explored. Not only need to distinguish unseen classes as in the 2D domain, 3D FSL is more challenging in terms of irregular structures, subtle inter-class differences, and high intra-class variances {when trained on a low number of data.} Moreover, different architectures and learning algorithms make it difficult to study the effectiveness of existing 2D FSL algorithms when migrating to the 3D domain. In this work, for the first time, we perform systematic and extensive investigations of directly applying recent 2D FSL works to 3D point cloud related backbone networks and thus suggest a strong learning baseline for few-shot 3D point cloud classification. Furthermore, we propose a new network, Point-cloud Correlation Interaction (PCIA), with three novel plug-and-play components called Salient-Part Fusion (SPF) module, Self-Channel Interaction Plus (SCI+) module, and Cross-Instance Fusion Plus (CIF+) module to obtain more representative embeddings and improve the feature distinction. These modules can be inserted into most FSL algorithms with minor changes and significantly improve the performance. Experimental results on three benchmark datasets, ModelNet40-FS, ShapeNet70-FS, and ScanObjectNN-FS, demonstrate that our method achieves state-of-the-art performance for the 3D FSL task. Code and datasets are available at https://github.com/cgye96/A_Closer_Look_At_3DFSL.

Sijin Chen, Xin Chen, Chi Zhang et al.

Recent advances in Large Multimodal Models (LMM) have made it possible for various applications in human-machine interactions. However, developing LMMs that can comprehend, reason, and plan in complex and diverse 3D environments remains a challenging topic, especially considering the demand for understanding permutation-invariant point cloud 3D representations of the 3D scene. Existing works seek help from multi-view images, and project 2D features to 3D space as 3D scene representations. This, however, leads to huge computational overhead and performance degradation. In this paper, we present LL3DA, a Large Language 3D Assistant that takes point cloud as direct input and respond to both textual-instructions and visual-prompts. This help LMMs better comprehend human interactions and further help to remove the ambiguities in cluttered 3D scenes. Experiments show that LL3DA achieves remarkable results, and surpasses various 3D vision-language models on both 3D Dense Captioning and 3D Question Answering.

Ke Xu, Jiangtao Wang, Hongyuan Zhu et al.

Recently, with the advancement of the Internet of Things (IoT), WiFi CSI-based HAR has gained increasing attention from academic and industry communities. By integrating the deep learning technology with CSI-based HAR, researchers achieve state-of-the-art performance without the need of expert knowledge. However, the scarcity of labeled CSI data remains the most prominent challenge when applying deep learning models in the context of CSI-based HAR due to the privacy and incomprehensibility of CSI-based HAR data. On the other hand, SSL has emerged as a promising approach for learning meaningful representations from data without heavy reliance on labeled examples. Therefore, considerable efforts have been made to address the challenge of insufficient data in deep learning by leveraging SSL algorithms. In this paper, we undertake a comprehensive inventory and analysis of the potential held by different categories of SSL algorithms, including those that have been previously studied and those that have not yet been explored, within the field. We provide an in-depth investigation of SSL algorithms in the context of WiFi CSI-based HAR. We evaluate four categories of SSL algorithms using three publicly available CSI HAR datasets, each encompassing different tasks and environmental settings. To ensure relevance to real-world applications, we design performance metrics that align with specific requirements. Furthermore, our experimental findings uncover several limitations and blind spots in existing work, highlighting the barriers that need to be addressed before SSL can be effectively deployed in real-world WiFi-based HAR applications. Our results also serve as a practical guideline for industry practitioners and provide valuable insights for future research endeavors in this field.

Tianfu Wang, Max Xiong, Jianxun Lian et al.

Social skills such as negotiation and leadership are crucial for personal and professional success in today's interconnected world. However, scalable and effective training remains a significant challenge due to the scarcity of expert coaching. In this paper, we introduce SocialCoach, a holistic LLM-powered agentic tutoring system for personalized social skill development at scale. First, SocialCoach automatically constructs a pedagogically-grounded, theory-to-practice knowledge corpus from diverse expert sources, leveraging a multi-agent pipeline. Second, to personalize the learning journey, it employs an adaptive practice scheduling module that follows a prescription-retrieval-adaptation process. To maximize the long-term learning experience while overcoming the cold-start problem, this policy is optimized within a learner simulation environment through reinforcement learning. Finally, SocialCoach integrates immersive, goal-driven practice, causality-driven proficiency assessment and knowledge-grounded, reflective tutoring to help address the knowing-doing gap. We deploy it in our product, EQoach, and conduct extensive experiments. The results show that SocialCoach improves simulated pathway quality and judge-rated tutoring quality over baseline approaches, while early user feedback indicates strong perceived engagement and usefulness. These findings suggest a practical architecture for personalized and gamified pedagogical platforms on soft skill learning.

Mingsheng Li, Xin Chen, Chi Zhang et al.

Recently, 3D understanding has become popular to facilitate autonomous agents to perform further decisionmaking. However, existing 3D datasets and methods are often limited to specific tasks. On the other hand, recent progress in Large Language Models (LLMs) and Multimodal Language Models (MLMs) have demonstrated exceptional general language and imagery tasking performance. Therefore, it is interesting to unlock MLM's potential to be 3D generalist for wider tasks. However, current MLMs' research has been less focused on 3D tasks due to a lack of large-scale 3D instruction-following datasets. In this work, we introduce a comprehensive 3D instructionfollowing dataset called M3DBench, which possesses the following characteristics: 1) It supports general multimodal instructions interleaved with text, images, 3D objects, and other visual prompts. 2) It unifies diverse 3D tasks at both region and scene levels, covering a variety of fundamental abilities in real-world 3D environments. 3) It is a large-scale 3D instruction-following dataset with over 320k instruction-response pairs. Furthermore, we establish a new benchmark for assessing the performance of large models in understanding multi-modal 3D prompts. Extensive experiments demonstrate the effectiveness of our dataset and baseline, supporting general 3D-centric tasks, which can inspire future research.

Ke Xu, Zhiyong Zheng, Hongyuan Zhu et al.

Wi-Fi sensing systems are severely hindered by cross domain problem when deployed in unseen real-world environments. Existing methods typically design separate frameworks for either domain adaptation or domain generalization, often relying on extensive labeled data. Existing methods that designed for domain generalization is often relying on extensive labeled data. However, real-world scenarios are far more complex, where the deployed model must be capable of handling generalization under limited labeled source data. To this end, we propose UniCrossFi, a unified framework designed to mitigate performance drop in CSI-based sensing across diverse deployment settings. Our framework not only extends conventional Domain Generalization (DG) to a more practical Semi-Supervised Domain Generalization (SSDG) setting, where only partially labeled source data are available, but also introduces a physics-informed data augmentation strategy, Antenna Response Consistency (ARC). ARC mitigates the risk of learning superficial shortcuts by exploiting the intrinsic spatial diversity of multi-antenna systems, treating signals from different antennas as naturally augmented views of the same event. In addition, we design a Unified Contrastive Objective to prevent conventional contrastive learning from pushing apart samples from different domains that share the same class. We conduct extensive experiments on the public Widar and CSIDA datasets. The results demonstrate that UniCrossFi consistently establishes a new state-of-the-art, significantly outperforming existing methods across all unsupervised domain adaptation, DG, and SSDG benchmarks. UniCrossFi provides a principled and practical solution to the domain shift challenge, advancing the feasibility of robust, real-world Wi-Fi sensing systems that can operate effectively with limited labeled data.

Zhijing Wan, Zhixiang Wang, Yuran Wang et al.

Coreset selection seeks to choose a subset of crucial training samples for efficient learning. It has gained traction in deep learning, particularly with the surge in training dataset sizes. Sample selection hinges on two main aspects: a sample's representation in enhancing performance and the role of sample diversity in averting overfitting. Existing methods typically measure both the representation and diversity of data based on similarity metrics, such as L2-norm. They have capably tackled representation via distribution matching guided by the similarities of features, gradients, or other information between data. However, the results of effectively diverse sample selection are mired in sub-optimality. This is because the similarity metrics usually simply aggregate dimension similarities without acknowledging disparities among the dimensions that significantly contribute to the final similarity. As a result, they fall short of adequately capturing diversity. To address this, we propose a feature-based diversity constraint, compelling the chosen subset to exhibit maximum diversity. Our key lies in the introduction of a novel Contributing Dimension Structure (CDS) metric. Different from similarity metrics that measure the overall similarity of high-dimensional features, our CDS metric considers not only the reduction of redundancy in feature dimensions, but also the difference between dimensions that contribute significantly to the final similarity. We reveal that existing methods tend to favor samples with similar CDS, leading to a reduced variety of CDS types within the coreset and subsequently hindering model performance. In response, we enhance the performance of five classical selection methods by integrating the CDS constraint. Our experiments on three datasets demonstrate the general effectiveness of the proposed method in boosting existing methods.

Shulin Tian, Ruiqi Wang, Hongming Guo et al.

We introduce Ego-R1, a novel framework for reasoning over ultra-long (i.e., in days and weeks) egocentric videos, which leverages a structured Chain-of-Tool-Thought (CoTT) process, orchestrated by an Ego-R1 Agent trained via reinforcement learning (RL). Inspired by human problem-solving strategies, CoTT decomposes complex reasoning into modular steps, with the RL agent invoking specific tools, one per step, to iteratively and collaboratively answer sub-questions tackling such tasks as temporal retrieval and multi-modal understanding. We design a two-stage training paradigm involving supervised finetuning (SFT) of a pretrained language model using CoTT data and RL to enable our agent to dynamically propose step-by-step tools for long-range reasoning. To facilitate training, we construct a dataset called Ego-R1 Data, which consists of Ego-CoTT-25K for SFT and Ego-QA-4.4K for RL. Furthermore, our Ego-R1 agent is evaluated on a newly curated week-long video QA benchmark, Ego-R1 Bench, which contains human-verified QA pairs from hybrid sources. Extensive results demonstrate that the dynamic, tool-augmented chain-of-thought reasoning by our Ego-R1 Agent can effectively tackle the unique challenges of understanding ultra-long egocentric videos, significantly extending the time coverage from few hours to a week.

Huaiyuan Qin, Muli Yang, Siyuan Hu et al.

Self-supervised learning (SSL) conventionally relies on the instance consistency paradigm, assuming that different views of the same image can be treated as positive pairs. However, this assumption breaks down for non-iconic data, where different views may contain distinct objects or semantic information. In this paper, we investigate the effectiveness of SSL when instance consistency is not guaranteed. Through extensive ablation studies, we demonstrate that SSL can still learn meaningful representations even when positive pairs lack strict instance consistency. Furthermore, our analysis further reveals that increasing view diversity, by enforcing zero overlapping or using smaller crop scales, can enhance downstream performance on classification and dense prediction tasks. However, excessive diversity is found to reduce effectiveness, suggesting an optimal range for view diversity. To quantify this, we adopt the Earth Mover's Distance (EMD) as an estimator to measure mutual information between views, finding that moderate EMD values correlate with improved SSL learning, providing insights for future SSL framework design. We validate our findings across a range of settings, highlighting their robustness and applicability on diverse data sources.

En Yen Puang, Hao Zhang, Hongyuan Zhu et al.

In this paper we present SA-CNN, a hierarchical and lightweight self-attention based encoding and decoding architecture for representation learning of point cloud data. The proposed SA-CNN introduces convolution and transposed convolution stacks to capture and generate contextual information among unordered 3D points. Following conventional hierarchical pipeline, the encoding process extracts feature in local-to-global manner, while the decoding process generates feature and point cloud in coarse-to-fine, multi-resolution stages. We demonstrate that SA-CNN is capable of a wide range of applications, namely classification, part segmentation, reconstruction, shape retrieval, and unsupervised classification. While achieving the state-of-the-art or comparable performance in the benchmarks, SA-CNN maintains its model complexity several order of magnitude lower than the others. In term of qualitative results, we visualize the multi-stage point cloud reconstructions and latent walks on rigid objects as well as deformable non-rigid human and robot models.

Yongbin Liao, Hongyuan Zhu, Yanggang Zhang et al.

Point cloud instance segmentation has achieved huge progress with the emergence of deep learning. However, these methods are usually data-hungry with expensive and time-consuming dense point cloud annotations. To alleviate the annotation cost, unlabeled or weakly labeled data is still less explored in the task. In this paper, we introduce the first semi-supervised point cloud instance segmentation framework (SPIB) using both labeled and unlabelled bounding boxes as supervision. To be specific, our SPIB architecture involves a two-stage learning procedure. For stage one, a bounding box proposal generation network is trained under a semi-supervised setting with perturbation consistency regularization (SPCR). The regularization works by enforcing an invariance of the bounding box predictions over different perturbations applied to the input point clouds, to provide self-supervision for network learning. For stage two, the bounding box proposals with SPCR are grouped into some subsets, and the instance masks are mined inside each subset with a novel semantic propagation module and a property consistency graph module. Moreover, we introduce a novel occupancy ratio guided refinement module to refine the instance masks. Extensive experiments on the challenging ScanNet v2 dataset demonstrate our method can achieve competitive performance compared with the recent fully-supervised methods.

Jiafei Duan, Samson Yu, Hui Li Tan et al.

There has been an emerging paradigm shift from the era of "internet AI" to "embodied AI", where AI algorithms and agents no longer learn from datasets of images, videos or text curated primarily from the internet. Instead, they learn through interactions with their environments from an egocentric perception similar to humans. Consequently, there has been substantial growth in the demand for embodied AI simulators to support various embodied AI research tasks. This growing interest in embodied AI is beneficial to the greater pursuit of Artificial General Intelligence (AGI), but there has not been a contemporary and comprehensive survey of this field. This paper aims to provide an encyclopedic survey for the field of embodied AI, from its simulators to its research. By evaluating nine current embodied AI simulators with our proposed seven features, this paper aims to understand the simulators in their provision for use in embodied AI research and their limitations. Lastly, this paper surveys the three main research tasks in embodied AI -- visual exploration, visual navigation and embodied question answering (QA), covering the state-of-the-art approaches, evaluation metrics and datasets. Finally, with the new insights revealed through surveying the field, the paper will provide suggestions for simulator-for-task selections and recommendations for the future directions of the field.

Tianying Wang, Hao Zhang, Wei Qi Toh et al.

Learning-based methods have been used to pro-gram robotic tasks in recent years. However, extensive training is usually required not only for the initial task learning but also for generalizing the learned model to the same task but in different environments. In this paper, we propose a novel Deep Reinforcement Learning algorithm for efficient task generalization and environment adaptation in the robotic task learning problem. The proposed method is able to efficiently generalize the previously learned task by model fusion to solve the environment adaptation problem. The proposed Deep Model Fusion (DMF) method reuses and combines the previously trained model to improve the learning efficiency and results.Besides, we also introduce a Multi-objective Guided Reward(MGR) shaping technique to further improve training efficiency.The proposed method was benchmarked with previous methods in various environments to validate its effectiveness.

Yi Cheng, Hongyuan Zhu, Ying Sun et al.

6D object pose estimation is widely applied in robotic tasks such as grasping and manipulation. Prior methods using RGB-only images are vulnerable to heavy occlusion and poor illumination, so it is important to complement them with depth information. However, existing methods using RGB-D data cannot adequately exploit consistent and complementary information between RGB and depth modalities. In this paper, we present a novel method to effectively consider the correlation within and across both modalities with attention mechanism to learn discriminative and compact multi-modal features. Then, effective fusion strategies for intra- and inter-correlation modules are explored to ensure efficient information flow between RGB and depth. To our best knowledge, this is the first work to explore effective intra- and inter-modality fusion in 6D pose estimation. The experimental results show that our method can achieve the state-of-the-art performance on LineMOD and YCB-Video dataset. We also demonstrate that the proposed method can benefit a real-world robot grasping task by providing accurate object pose estimation.

Zhao Kang, Honghui Xu, Boyu Wang et al.

Graph-based clustering has shown promising performance in many tasks. A key step of graph-based approach is the similarity graph construction. In general, learning graph in kernel space can enhance clustering accuracy due to the incorporation of nonlinearity. However, most existing kernel-based graph learning mechanisms is not similarity-preserving, hence leads to sub-optimal performance. To overcome this drawback, we propose a more discriminative graph learning method which can preserve the pairwise similarities between samples in an adaptive manner for the first time. Specifically, we require the learned graph be close to a kernel matrix, which serves as a measure of similarity in raw data. Moreover, the structure is adaptively tuned so that the number of connected components of the graph is exactly equal to the number of clusters. Finally, our method unifies clustering and graph learning which can directly obtain cluster indicators from the graph itself without performing further clustering step. The effectiveness of this approach is examined on both single and multiple kernel learning scenarios in several datasets.

Changgong Zhang, Fangneng Zhan, Hongyuan Zhu et al.

A large amount of annotated training images is critical for training accurate and robust deep network models but the collection of a large amount of annotated training images is often time-consuming and costly. Image synthesis alleviates this constraint by generating annotated training images automatically by machines which has attracted increasing interest in the recent deep learning research. We develop an innovative image synthesis technique that composes annotated training images by realistically embedding foreground objects of interest (OOI) into background images. The proposed technique consists of two key components that in principle boost the usefulness of the synthesized images in deep network training. The first is context-aware semantic coherence which ensures that the OOI are placed around semantically coherent regions within the background image. The second is harmonious appearance adaptation which ensures that the embedded OOI are agreeable to the surrounding background from both geometry alignment and appearance realism. The proposed technique has been evaluated over two related but very different computer vision challenges, namely, scene text detection and scene text recognition. Experiments over a number of public datasets demonstrate the effectiveness of our proposed image synthesis technique - the use of our synthesized images in deep network training is capable of achieving similar or even better scene text detection and scene text recognition performance as compared with using real images.

Fangneng Zhan, Hongyuan Zhu, Shijian Lu

Recent advances in generative adversarial networks (GANs) have shown great potentials in realistic image synthesis whereas most existing works address synthesis realism in either appearance space or geometry space but few in both. This paper presents an innovative Spatial Fusion GAN (SF-GAN) that combines a geometry synthesizer and an appearance synthesizer to achieve synthesis realism in both geometry and appearance spaces. The geometry synthesizer learns contextual geometries of background images and transforms and places foreground objects into the background images unanimously. The appearance synthesizer adjusts the color, brightness and styles of the foreground objects and embeds them into background images harmoniously, where a guided filter is introduced for detail preserving. The two synthesizers are inter-connected as mutual references which can be trained end-to-end without supervision. The SF-GAN has been evaluated in two tasks: (1) realistic scene text image synthesis for training better recognition models; (2) glass and hat wearing for realistic matching glasses and hats with real portraits. Qualitative and quantitative comparisons with the state-of-the-art demonstrate the superiority of the proposed SF-GAN.

Anran Wang, Anh Tuan Luu, Chuan-Sheng Foo et al.

Answering questions according to multi-modal context is a challenging problem as it requires a deep integration of different data sources. Existing approaches only employ partial interactions among data sources in one attention hop. In this paper, we present the Holistic Multi-modal Memory Network (HMMN) framework which fully considers the interactions between different input sources (multi-modal context, question) in each hop. In addition, it takes answer choices into consideration during the context retrieval stage. Therefore, the proposed framework effectively integrates multi-modal context, question, and answer information, which leads to more informative context retrieved for question answering. Our HMMN framework achieves state-of-the-art accuracy on MovieQA dataset. Extensive ablation studies show the importance of holistic reasoning and contributions of different attention strategies.

Zhe Wang, Kingsley Kuan, Mathieu Ravaut et al.

The YouTube-8M video classification challenge requires teams to classify 0.7 million videos into one or more of 4,716 classes. In this Kaggle competition, we placed in the top 3% out of 650 participants using released video and audio features. Beyond that, we extend the original competition by including text information in the classification, making this a truly multi-modal approach with vision, audio and text. The newly introduced text data is termed as YouTube-8M-Text. We present a classification framework for the joint use of text, visual and audio features, and conduct an extensive set of experiments to quantify the benefit that this additional mode brings. The inclusion of text yields state-of-the-art results, e.g. 86.7% GAP on the YouTube-8M-Text validation dataset.

Hongyuan Zhu, Shijian Lu, Jianfei Cai et al.

Object proposal has become a popular paradigm to replace exhaustive sliding window search in current top-performing methods in PASCAL VOC and ImageNet. Recently, Hosang et al. conduct the first unified study of existing methods' in terms of various image-level degradations. On the other hand, the vital question "what object-level characteristics really affect existing methods' performance?" is not yet answered. Inspired by Hoiem et al.'s work in categorical object detection, this paper conducts the first meta-analysis of various object-level characteristics' impact on state-of-the-art object proposal methods. Specifically, we examine the effects of object size, aspect ratio, iconic view, color contrast, shape regularity and texture. We also analyse existing methods' localization accuracy and latency for various PASCAL VOC object classes. Our study reveals the limitations of existing methods in terms of non-iconic view, small object size, low color contrast, shape regularity etc. Based on our observations, lessons are also learned and shared with respect to the selection of existing object proposal technologies as well as the design of the future ones.

Hongyuan Zhu, Fanman Meng, Jianfei Cai et al.

Image segmentation refers to the process to divide an image into nonoverlapping meaningful regions according to human perception, which has become a classic topic since the early ages of computer vision. A lot of research has been conducted and has resulted in many applications. However, while many segmentation algorithms exist, yet there are only a few sparse and outdated summarizations available, an overview of the recent achievements and issues is lacking. We aim to provide a comprehensive review of the recent progress in this field. Covering 180 publications, we give an overview of broad areas of segmentation topics including not only the classic bottom-up approaches, but also the recent development in superpixel, interactive methods, object proposals, semantic image parsing and image cosegmentation. In addition, we also review the existing influential datasets and evaluation metrics. Finally, we suggest some design flavors and research directions for future research in image segmentation.