Yanjun Wang

Xiaoyang Huang, Jiancheng Yang, Yanjun Wang et al.

3D shape analysis has been widely explored in the era of deep learning. Numerous models have been developed for various 3D data representation formats, e.g., MeshCNN for meshes, PointNet for point clouds and VoxNet for voxels. In this study, we present Representation-Agnostic Shape Fields (RASF), a generalizable and computation-efficient shape embedding module for 3D deep learning. RASF is implemented with a learnable 3D grid with multiple channels to store local geometry. Based on RASF, shape embeddings for various 3D shape representations (point clouds, meshes and voxels) are retrieved by coordinate indexing. While there are multiple ways to optimize the learnable parameters of RASF, we provide two effective schemes among all in this paper for RASF pre-training: shape reconstruction and normal estimation. Once trained, RASF becomes a plug-and-play performance booster with negligible cost. Extensive experiments on diverse 3D representation formats, networks and applications, validate the universal effectiveness of the proposed RASF. Code and pre-trained models are publicly available https://github.com/seanywang0408/RASF

Xiaoyang Huang, Yanjun Wang, Yang Liu et al.

Spatial audio, which focuses on immersive 3D sound rendering, is widely applied in the acoustic industry. One of the key problems of current spatial audio rendering methods is the lack of personalization based on different anatomies of individuals, which is essential to produce accurate sound source positions. In this work, we address this problem from an interdisciplinary perspective. The rendering of spatial audio is strongly correlated with the 3D shape of human bodies, particularly ears. To this end, we propose to achieve personalized spatial audio by reconstructing 3D human ears with single-view images. First, to benchmark the ear reconstruction task, we introduce AudioEar3D, a high-quality 3D ear dataset consisting of 112 point cloud ear scans with RGB images. To self-supervisedly train a reconstruction model, we further collect a 2D ear dataset composed of 2,000 images, each one with manual annotation of occlusion and 55 landmarks, named AudioEar2D. To our knowledge, both datasets have the largest scale and best quality of their kinds for public use. Further, we propose AudioEarM, a reconstruction method guided by a depth estimation network that is trained on synthetic data, with two loss functions tailored for ear data. Lastly, to fill the gap between the vision and acoustics community, we develop a pipeline to integrate the reconstructed ear mesh with an off-the-shelf 3D human body and simulate a personalized Head-Related Transfer Function (HRTF), which is the core of spatial audio rendering. Code and data are publicly available at https://github.com/seanywang0408/AudioEar.

Zhongang Cai, Wanqi Yin, Ailing Zeng et al.

Expressive human pose and shape estimation (EHPS) unifies body, hands, and face motion capture with numerous applications. Despite encouraging progress, current state-of-the-art methods still depend largely on a confined set of training datasets. In this work, we investigate scaling up EHPS towards the first generalist foundation model (dubbed SMPLer-X), with up to ViT-Huge as the backbone and training with up to 4.5M instances from diverse data sources. With big data and the large model, SMPLer-X exhibits strong performance across diverse test benchmarks and excellent transferability to even unseen environments. 1) For the data scaling, we perform a systematic investigation on 32 EHPS datasets, including a wide range of scenarios that a model trained on any single dataset cannot handle. More importantly, capitalizing on insights obtained from the extensive benchmarking process, we optimize our training scheme and select datasets that lead to a significant leap in EHPS capabilities. 2) For the model scaling, we take advantage of vision transformers to study the scaling law of model sizes in EHPS. Moreover, our finetuning strategy turn SMPLer-X into specialist models, allowing them to achieve further performance boosts. Notably, our foundation model SMPLer-X consistently delivers state-of-the-art results on seven benchmarks such as AGORA (107.2 mm NMVE), UBody (57.4 mm PVE), EgoBody (63.6 mm PVE), and EHF (62.3 mm PVE without finetuning). Homepage: https://caizhongang.github.io/projects/SMPLer-X/

Wenjia Wang, Yongtao Ge, Haiyi Mei et al.

As it is hard to calibrate single-view RGB images in the wild, existing 3D human mesh reconstruction (3DHMR) methods either use a constant large focal length or estimate one based on the background environment context, which can not tackle the problem of the torso, limb, hand or face distortion caused by perspective camera projection when the camera is close to the human body. The naive focal length assumptions can harm this task with the incorrectly formulated projection matrices. To solve this, we propose Zolly, the first 3DHMR method focusing on perspective-distorted images. Our approach begins with analysing the reason for perspective distortion, which we find is mainly caused by the relative location of the human body to the camera center. We propose a new camera model and a novel 2D representation, termed distortion image, which describes the 2D dense distortion scale of the human body. We then estimate the distance from distortion scale features rather than environment context features. Afterwards, we integrate the distortion feature with image features to reconstruct the body mesh. To formulate the correct projection matrix and locate the human body position, we simultaneously use perspective and weak-perspective projection loss. Since existing datasets could not handle this task, we propose the first synthetic dataset PDHuman and extend two real-world datasets tailored for this task, all containing perspective-distorted human images. Extensive experiments show that Zolly outperforms existing state-of-the-art methods on both perspective-distorted datasets and the standard benchmark (3DPW).

Yanjun Wang, Qingping Sun, Wenjia Wang et al.

Human mesh reconstruction from a single image is challenging in the presence of occlusion, which can be caused by self, objects, or other humans. Existing methods either fail to separate human features accurately or lack proper supervision for feature completion. In this paper, we propose Dense Inpainting Human Mesh Recovery (DIMR), a two-stage method that leverages dense correspondence maps to handle occlusion. Our method utilizes a dense correspondence map to separate visible human features and completes human features on a structured UV map dense human with an attention-based feature completion module. We also design a feature inpainting training procedure that guides the network to learn from unoccluded features. We evaluate our method on several datasets and demonstrate its superior performance under heavily occluded scenarios compared to other methods. Extensive experiments show that our method obviously outperforms prior SOTA methods on heavily occluded images and achieves comparable results on the standard benchmarks (3DPW).

Wanqi Yin, Zhongang Cai, Ruisi Wang et al.

Expressive human pose and shape estimation (EHPS) unifies body, hands, and face motion capture with numerous applications. Despite encouraging progress, current state-of-the-art methods focus on training innovative architectural designs on confined datasets. In this work, we investigate the impact of scaling up EHPS towards a family of generalist foundation models. 1) For data scaling, we perform a systematic investigation on 40 EHPS datasets, encompassing a wide range of scenarios that a model trained on any single dataset cannot handle. More importantly, capitalizing on insights obtained from the extensive benchmarking process, we optimize our training scheme and select datasets that lead to a significant leap in EHPS capabilities. Ultimately, we achieve diminishing returns at 10M training instances from diverse data sources. 2) For model scaling, we take advantage of vision transformers (up to ViT-Huge as the backbone) to study the scaling law of model sizes in EHPS. To exclude the influence of algorithmic design, we base our experiments on two minimalist architectures: SMPLer-X, which consists of an intermediate step for hand and face localization, and SMPLest-X, an even simpler version that reduces the network to its bare essentials and highlights significant advances in the capture of articulated hands. With big data and the large model, the foundation models exhibit strong performance across diverse test benchmarks and excellent transferability to even unseen environments. Moreover, our finetuning strategy turns the generalist into specialist models, allowing them to achieve further performance boosts. Notably, our foundation models consistently deliver state-of-the-art results on seven benchmarks such as AGORA, UBody, EgoBody, and our proposed SynHand dataset for comprehensive hand evaluation. (Code is available at: https://github.com/wqyin/SMPLest-X).

Bing Wang, Ximing Li, Yanjun Wang et al.

Multimodal Misinformation Detection (MMD) refers to the task of detecting social media posts involving misinformation, where the post often contains text and image modalities. However, by observing the MMD posts, we hold that the text modality may be much more informative than the image modality because the text generally describes the whole event/story of the current post but the image often presents partial scenes only. Our preliminary empirical results indicate that the image modality exactly contributes less to MMD. Upon this idea, we propose a new MMD method named RETSIMD. Specifically, we suppose that each text can be divided into several segments, and each text segment describes a partial scene that can be presented by an image. Accordingly, we split the text into a sequence of segments, and feed these segments into a pre-trained text-to-image generator to augment a sequence of images. We further incorporate two auxiliary objectives concerning text-image and image-label mutual information, and further post-train the generator over an auxiliary text-to-image generation benchmark dataset. Additionally, we propose a graph structure by defining three heuristic relationships between images, and use a graph neural network to generate the fused features. Extensive empirical results validate the effectiveness of RETSIMD.

Zheng Li, Caili Guo, Xin Wang et al.

Image-Text Retrieval (ITR) is essentially a ranking problem. Given a query caption, the goal is to rank candidate images by relevance, from large to small. The current ITR datasets are constructed in a pairwise manner. Image-text pairs are annotated as positive or negative. Correspondingly, ITR models mainly use pairwise losses, such as triplet loss, to learn to rank. Pairwise-based ITR increases positive pair similarity while decreasing negative pair similarity indiscriminately. However, the relevance between dissimilar negative pairs is different. Pairwise annotations cannot reflect this difference in relevance. In the current datasets, pairwise annotations miss many correlations. There are many potential positive pairs among the pairs labeled as negative. Pairwise-based ITR can only rank positive samples before negative samples, but cannot rank negative samples by relevance. In this paper, we integrate listwise ranking into conventional pairwise-based ITR. Listwise ranking optimizes the entire ranking list based on relevance scores. Specifically, we first propose a Relevance Score Calculation (RSC) module to calculate the relevance score of the entire ranked list. Then we choose the ranking metric, Normalized Discounted Cumulative Gain (NDCG), as the optimization objective. We transform the non-differentiable NDCG into a differentiable listwise loss, named Smooth-NDCG (S-NDCG). Our listwise ranking approach can be plug-and-play integrated into current pairwise-based ITR models. Experiments on ITR benchmarks show that integrating listwise ranking can improve the performance of current ITR models and provide more user-friendly retrieval results. The code is available at https://github.com/AAA-Zheng/Listwise_ITR.

Qingping Sun, Yanjun Wang, Ailing Zeng et al.

Expressive human pose and shape estimation (a.k.a. 3D whole-body mesh recovery) involves the human body, hand, and expression estimation. Most existing methods have tackled this task in a two-stage manner, first detecting the human body part with an off-the-shelf detection model and inferring the different human body parts individually. Despite the impressive results achieved, these methods suffer from 1) loss of valuable contextual information via cropping, 2) introducing distractions, and 3) lacking inter-association among different persons and body parts, inevitably causing performance degradation, especially for crowded scenes. To address these issues, we introduce a novel all-in-one-stage framework, AiOS, for multiple expressive human pose and shape recovery without an additional human detection step. Specifically, our method is built upon DETR, which treats multi-person whole-body mesh recovery task as a progressive set prediction problem with various sequential detection. We devise the decoder tokens and extend them to our task. Specifically, we first employ a human token to probe a human location in the image and encode global features for each instance, which provides a coarse location for the later transformer block. Then, we introduce a joint-related token to probe the human joint in the image and encoder a fine-grained local feature, which collaborates with the global feature to regress the whole-body mesh. This straightforward but effective model outperforms previous state-of-the-art methods by a 9% reduction in NMVE on AGORA, a 30% reduction in PVE on EHF, a 10% reduction in PVE on ARCTIC, and a 3% reduction in PVE on EgoBody.

Shaochun Li, Yanjun Wang, Hengfan Cai et al.

Real-time and accurate information on fine-grained changes in crop cultivation is of great significance for crop growth monitoring, yield prediction and agricultural structure adjustment. Aiming at the problems of serious spectral confusion in visible high-resolution unmanned aerial vehicle (UAV) images of different phases, interference of large complex background and salt-and-pepper noise by existing semantic change detection (SCD) algorithms, in order to effectively extract deep image features of crops and meet the demand of agricultural practical engineering applications, this paper designs and proposes an agricultural geographic scene and parcel-scale constrained SCD framework for crops (AGSPNet). AGSPNet framework contains three parts: agricultural geographic scene (AGS) division module, parcel edge extraction module and crop SCD module. Meanwhile, we produce and introduce an UAV image SCD dataset (CSCD) dedicated to agricultural monitoring, encompassing multiple semantic variation types of crops in complex geographical scene. We conduct comparative experiments and accuracy evaluations in two test areas of this dataset, and the results show that the crop SCD results of AGSPNet consistently outperform other deep learning SCD models in terms of quantity and quality, with the evaluation metrics F1-score, kappa, OA, and mIoU obtaining improvements of 0.038, 0.021, 0.011 and 0.062, respectively, on average over the sub-optimal method. The method proposed in this paper can clearly detect the fine-grained change information of crop types in complex scenes, which can provide scientific and technical support for smart agriculture monitoring and management, food policy formulation and food security assurance.

Wenbo Hu, Yining Hong, Yanjun Wang et al.

Humans excel at performing complex tasks by leveraging long-term memory across temporal and spatial experiences. In contrast, current Large Language Models (LLMs) struggle to effectively plan and act in dynamic, multi-room 3D environments. We posit that part of this limitation is due to the lack of proper 3D spatial-temporal memory modeling in LLMs. To address this, we first introduce 3DMem-Bench, a comprehensive benchmark comprising over 26,000 trajectories and 2,892 embodied tasks, question-answering and captioning, designed to evaluate an agent's ability to reason over long-term memory in 3D environments. Second, we propose 3DLLM-Mem, a novel dynamic memory management and fusion model for embodied spatial-temporal reasoning and actions in LLMs. Our model uses working memory tokens, which represents current observations, as queries to selectively attend to and fuse the most useful spatial and temporal features from episodic memory, which stores past observations and interactions. Our approach allows the agent to focus on task-relevant information while maintaining memory efficiency in complex, long-horizon environments. Experimental results demonstrate that 3DLLM-Mem achieves state-of-the-art performance across various tasks, outperforming the strongest baselines by 16.5% in success rate on 3DMem-Bench's most challenging in-the-wild embodied tasks.

Jiasen Lu, Liangchen Song, Mingze Xu et al.

We present AToken, the first unified visual tokenizer that achieves both high-fidelity reconstruction and semantic understanding across images, videos, and 3D assets. Unlike existing tokenizers that specialize in either reconstruction or understanding for single modalities, AToken encodes these diverse visual inputs into a shared 4D latent space, unifying both tasks and modalities in a single framework. Specifically, we introduce a pure transformer architecture with 4D rotary position embeddings to process visual inputs of arbitrary resolutions and temporal durations. To ensure stable training, we introduce an adversarial-free training objective that combines perceptual and Gram matrix losses, achieving state-of-the-art reconstruction quality. By employing a progressive training curriculum, AToken gradually expands from single images, videos, and 3D, and supports both continuous and discrete latent tokens. AToken achieves 0.21 rFID with 82.2% ImageNet accuracy for images, 3.01 rFVD with 40.2% MSRVTT retrieval for videos, and 28.28 PSNR with 90.9% classification accuracy for 3D.. In downstream applications, AToken enables both visual generation tasks (e.g., image generation with continuous and discrete tokens, text-to-video generation, image-to-3D synthesis) and understanding tasks (e.g., multimodal LLMs), achieving competitive performance across all benchmarks. These results shed light on the next-generation multimodal AI systems built upon unified visual tokenization.

Yanjun Wang, Zixuan Li, Chuan Jiang et al.

When managing wide-area networks, network architects must decide how to balance multiple conflicting metrics, and ensure fair allocations to competing traffic while prioritizing critical traffic. The state of practice poses challenges since architects must precisely encode their intent into formal optimization models using abstract notions such as utility functions, and ad-hoc manually tuned knobs. In this paper, we present the first effort to synthesize optimal network designs with indeterminate objectives using an interactive program-synthesis-based approach. We make three contributions. First, we present comparative synthesis, an interactive synthesis framework which produces near-optimal programs (network designs) through two kinds of queries (Propose and Compare), without an objective explicitly given. Second, we develop the first learning algorithm for comparative synthesis in which a voting-guided learner picks the most informative query in each iteration. We present theoretical analysis of the convergence rate of the algorithm. Third, we implemented Net10Q, a system based on our approach, and demonstrate its effectiveness on four real-world network case studies using black-box oracles and simulation experiments, as well as a pilot user study comprising network researchers and practitioners. Both theoretical and experimental results show the promise of our approach.

Honghong Zhou, Caili Guo, Hao Zhang et al.

In spite of the great progress in human motion prediction, it is still a challenging task to predict those aperiodic and complicated motions. We believe that to capture the correlations among human body components is the key to understand the human motion. In this paper, we propose a novel multiscale graph convolution network (MGCN) to address this problem. Firstly, we design an adaptive multiscale interactional encoding module (MIEM) which is composed of two sub modules: scale transformation module and scale interaction module to learn the human body correlations. Secondly, we apply a coarse-to-fine decoding strategy to decode the motions sequentially. We evaluate our approach on two standard benchmark datasets for human motion prediction: Human3.6M and CMU motion capture dataset. The experiments show that the proposed approach achieves the state-of-the-art performance for both short-term and long-term prediction especially in those complicated action category.

Weizun Zhao, Lishuai Li, Sameer Alam et al.

Safety is a top priority for civil aviation. New anomaly detection methods, primarily clustering methods, have been developed to monitor pilot operations and detect any risks from such flight data. However, all existing anomaly detection methods are offlline learning - the models are trained once using historical data and used for all future predictions. In practice, new flight data are accumulated continuously and analyzed every month at airlines. Clustering such dynamically growing data is challenging for an offlline method because it is memory and time intensive to re-train the model every time new data come in. If the model is not re-trained, false alarms or missed detections may increase since the model cannot reflect changes in data patterns. To address this problem, we propose a novel incremental anomaly detection method based on Gaussian Mixture Model (GMM) to identify common patterns and detect outliers in flight operations from digital flight data. It is a probabilistic clustering model of flight operations that can incrementally update its clusters based on new data rather than to re-cluster all data from scratch. It trains an initial GMM model based on historical offlline data. Then, it continuously adapts to new incoming data points via an expectation-maximization (EM) algorithm. To track changes in flight operation patterns, only model parameters need to be saved. The proposed method was tested on three sets of simulation data and two sets of real-world flight data. Compared with the traditional offline GMM method, the proposed method can generate similar clustering results with significantly reduced processing time (57 % - 99 % time reduction in testing sets) and memory usage (91 % - 95 % memory usage reduction in testing sets). Preliminary results indicate that the incremental learning scheme is effective in dealing with dynamically growing data in flight data analytics.