Shijian Lu

Jingyi Zhang, Jiaxing Huang, Sheng Jin et al.

Most visual recognition studies rely heavily on crowd-labelled data in deep neural networks (DNNs) training, and they usually train a DNN for each single visual recognition task, leading to a laborious and time-consuming visual recognition paradigm. To address the two challenges, Vision-Language Models (VLMs) have been intensively investigated recently, which learns rich vision-language correlation from web-scale image-text pairs that are almost infinitely available on the Internet and enables zero-shot predictions on various visual recognition tasks with a single VLM. This paper provides a systematic review of visual language models for various visual recognition tasks, including: (1) the background that introduces the development of visual recognition paradigms; (2) the foundations of VLM that summarize the widely-adopted network architectures, pre-training objectives, and downstream tasks; (3) the widely-adopted datasets in VLM pre-training and evaluations; (4) the review and categorization of existing VLM pre-training methods, VLM transfer learning methods, and VLM knowledge distillation methods; (5) the benchmarking, analysis and discussion of the reviewed methods; (6) several research challenges and potential research directions that could be pursued in the future VLM studies for visual recognition. A project associated with this survey has been created at https://github.com/jingyi0000/VLM_survey.

Gongjie Zhang, Zhipeng Luo, Kaiwen Cui et al.

Few-shot object detection has been extensively investigated by incorporating meta-learning into region-based detection frameworks. Despite its success, the said paradigm is still constrained by several factors, such as (i) low-quality region proposals for novel classes and (ii) negligence of the inter-class correlation among different classes. Such limitations hinder the generalization of base-class knowledge for the detection of novel-class objects. In this work, we design Meta-DETR, which (i) is the first image-level few-shot detector, and (ii) introduces a novel inter-class correlational meta-learning strategy to capture and leverage the correlation among different classes for robust and accurate few-shot object detection. Meta-DETR works entirely at image level without any region proposals, which circumvents the constraint of inaccurate proposals in prevalent few-shot detection frameworks. In addition, the introduced correlational meta-learning enables Meta-DETR to simultaneously attend to multiple support classes within a single feedforward, which allows to capture the inter-class correlation among different classes, thus significantly reducing the misclassification over similar classes and enhancing knowledge generalization to novel classes. Experiments over multiple few-shot object detection benchmarks show that the proposed Meta-DETR outperforms state-of-the-art methods by large margins. The implementation codes are available at https://github.com/ZhangGongjie/Meta-DETR.

Gongjie Zhang, Zhipeng Luo, Yingchen Yu et al.

The recently developed DEtection TRansformer (DETR) establishes a new object detection paradigm by eliminating a series of hand-crafted components. However, DETR suffers from extremely slow convergence, which increases the training cost significantly. We observe that the slow convergence is largely attributed to the complication in matching object queries with target features in different feature embedding spaces. This paper presents SAM-DETR, a Semantic-Aligned-Matching DETR that greatly accelerates DETR's convergence without sacrificing its accuracy. SAM-DETR addresses the convergence issue from two perspectives. First, it projects object queries into the same embedding space as encoded image features, where the matching can be accomplished efficiently with aligned semantics. Second, it explicitly searches salient points with the most discriminative features for semantic-aligned matching, which further speeds up the convergence and boosts detection accuracy as well. Being like a plug and play, SAM-DETR complements existing convergence solutions well yet only introduces slight computational overhead. Extensive experiments show that the proposed SAM-DETR achieves superior convergence as well as competitive detection accuracy. The implementation codes are available at https://github.com/ZhangGongjie/SAM-DETR.

Gongjie Zhang, Zhipeng Luo, Jiaxing Huang et al.

The recently proposed DEtection TRansformer (DETR) has established a fully end-to-end paradigm for object detection. However, DETR suffers from slow training convergence, which hinders its applicability to various detection tasks. We observe that DETR's slow convergence is largely attributed to the difficulty in matching object queries to relevant regions due to the unaligned semantics between object queries and encoded image features. With this observation, we design Semantic-Aligned-Matching DETR++ (SAM-DETR++) to accelerate DETR's convergence and improve detection performance. The core of SAM-DETR++ is a plug-and-play module that projects object queries and encoded image features into the same feature embedding space, where each object query can be easily matched to relevant regions with similar semantics. Besides, SAM-DETR++ searches for multiple representative keypoints and exploits their features for semantic-aligned matching with enhanced representation capacity. Furthermore, SAM-DETR++ can effectively fuse multi-scale features in a coarse-to-fine manner on the basis of the designed semantic-aligned matching. Extensive experiments show that the proposed SAM-DETR++ achieves superior convergence speed and competitive detection accuracy. Additionally, as a plug-and-play method, SAM-DETR++ can complement existing DETR convergence solutions with even better performance, achieving 44.8% AP with merely 12 training epochs and 49.1% AP with 50 training epochs on COCO val2017 with ResNet-50. Codes are available at https://github.com/ZhangGongjie/SAM-DETR .

Aoran Xiao, Jiaxing Huang, Weihao Xuan et al.

Robust point cloud parsing under all-weather conditions is crucial to level-5 autonomy in autonomous driving. However, how to learn a universal 3D semantic segmentation (3DSS) model is largely neglected as most existing benchmarks are dominated by point clouds captured under normal weather. We introduce SemanticSTF, an adverse-weather point cloud dataset that provides dense point-level annotations and allows to study 3DSS under various adverse weather conditions. We study all-weather 3DSS modeling under two setups: 1) domain adaptive 3DSS that adapts from normal-weather data to adverse-weather data; 2) domain generalizable 3DSS that learns all-weather 3DSS models from normal-weather data. Our studies reveal the challenge while existing 3DSS methods encounter adverse-weather data, showing the great value of SemanticSTF in steering the future endeavor along this very meaningful research direction. In addition, we design a domain randomization technique that alternatively randomizes the geometry styles of point clouds and aggregates their embeddings, ultimately leading to a generalizable model that can improve 3DSS under various adverse weather effectively. The SemanticSTF and related codes are available at \url{https://github.com/xiaoaoran/SemanticSTF}.

Fangneng Zhan, Jiahui Zhang, Yingchen Yu et al.

Perceiving the similarity between images has been a long-standing and fundamental problem underlying various visual generation tasks. Predominant approaches measure the inter-image distance by computing pointwise absolute deviations, which tends to estimate the median of instance distributions and leads to blurs and artifacts in the generated images. This paper presents MoNCE, a versatile metric that introduces image contrast to learn a calibrated metric for the perception of multifaceted inter-image distances. Unlike vanilla contrast which indiscriminately pushes negative samples from the anchor regardless of their similarity, we propose to re-weight the pushing force of negative samples adaptively according to their similarity to the anchor, which facilitates the contrastive learning from informative negative samples. Since multiple patch-level contrastive objectives are involved in image distance measurement, we introduce optimal transport in MoNCE to modulate the pushing force of negative samples collaboratively across multiple contrastive objectives. Extensive experiments over multiple image translation tasks show that the proposed MoNCE outperforms various prevailing metrics substantially. The code is available at https://github.com/fnzhan/MoNCE.

Zuhao Yang, Fangneng Zhan, Kunhao Liu et al.

The advancement of visual intelligence is intrinsically tethered to the availability of large-scale data. In parallel, generative Artificial Intelligence (AI) has unlocked the potential to create synthetic images that closely resemble real-world photographs. This prompts a compelling inquiry: how much visual intelligence could benefit from the advance of generative AI? This paper explores the innovative concept of harnessing these AI-generated images as new data sources, reshaping traditional modeling paradigms in visual intelligence. In contrast to real data, AI-generated data exhibit remarkable advantages, including unmatched abundance and scalability, the rapid generation of vast datasets, and the effortless simulation of edge cases. Built on the success of generative AI models, we examine the potential of their generated data in a range of applications, from training machine learning models to simulating scenarios for computational modeling, testing, and validation. We probe the technological foundations that support this groundbreaking use of generative AI, engaging in an in-depth discussion on the ethical, legal, and practical considerations that accompany this transformative paradigm shift. Through an exhaustive survey of current technologies and applications, this paper presents a comprehensive view of the synthetic era in visual intelligence. A project associated with this paper can be found at https://github.com/mwxely/AIGS .

Kunhao Liu, Fangneng Zhan, Yiwen Chen et al.

3D style transfer aims to render stylized novel views of a 3D scene with multi-view consistency. However, most existing work suffers from a three-way dilemma over accurate geometry reconstruction, high-quality stylization, and being generalizable to arbitrary new styles. We propose StyleRF (Style Radiance Fields), an innovative 3D style transfer technique that resolves the three-way dilemma by performing style transformation within the feature space of a radiance field. StyleRF employs an explicit grid of high-level features to represent 3D scenes, with which high-fidelity geometry can be reliably restored via volume rendering. In addition, it transforms the grid features according to the reference style which directly leads to high-quality zero-shot style transfer. StyleRF consists of two innovative designs. The first is sampling-invariant content transformation that makes the transformation invariant to the holistic statistics of the sampled 3D points and accordingly ensures multi-view consistency. The second is deferred style transformation of 2D feature maps which is equivalent to the transformation of 3D points but greatly reduces memory footprint without degrading multi-view consistency. Extensive experiments show that StyleRF achieves superior 3D stylization quality with precise geometry reconstruction and it can generalize to various new styles in a zero-shot manner.

Jiahui Zhang, Fangneng Zhan, Christian Theobalt et al.

Quantizing images into discrete representations has been a fundamental problem in unified generative modeling. Predominant approaches learn the discrete representation either in a deterministic manner by selecting the best-matching token or in a stochastic manner by sampling from a predicted distribution. However, deterministic quantization suffers from severe codebook collapse and misalignment with inference stage while stochastic quantization suffers from low codebook utilization and perturbed reconstruction objective. This paper presents a regularized vector quantization framework that allows to mitigate above issues effectively by applying regularization from two perspectives. The first is a prior distribution regularization which measures the discrepancy between a prior token distribution and the predicted token distribution to avoid codebook collapse and low codebook utilization. The second is a stochastic mask regularization that introduces stochasticity during quantization to strike a good balance between inference stage misalignment and unperturbed reconstruction objective. In addition, we design a probabilistic contrastive loss which serves as a calibrated metric to further mitigate the perturbed reconstruction objective. Extensive experiments show that the proposed quantization framework outperforms prevailing vector quantization methods consistently across different generative models including auto-regressive models and diffusion models.

Chuhui Xue, Wenqing Zhang, Yu Hao et al.

Recently, Vision-Language Pre-training (VLP) techniques have greatly benefited various vision-language tasks by jointly learning visual and textual representations, which intuitively helps in Optical Character Recognition (OCR) tasks due to the rich visual and textual information in scene text images. However, these methods cannot well cope with OCR tasks because of the difficulty in both instance-level text encoding and image-text pair acquisition (i.e. images and captured texts in them). This paper presents a weakly supervised pre-training method, oCLIP, which can acquire effective scene text representations by jointly learning and aligning visual and textual information. Our network consists of an image encoder and a character-aware text encoder that extract visual and textual features, respectively, as well as a visual-textual decoder that models the interaction among textual and visual features for learning effective scene text representations. With the learning of textual features, the pre-trained model can attend texts in images well with character awareness. Besides, these designs enable the learning from weakly annotated texts (i.e. partial texts in images without text bounding boxes) which mitigates the data annotation constraint greatly. Experiments over the weakly annotated images in ICDAR2019-LSVT show that our pre-trained model improves F-score by +2.5\% and +4.8\% while transferring its weights to other text detection and spotting networks, respectively. In addition, the proposed method outperforms existing pre-training techniques consistently across multiple public datasets (e.g., +3.2\% and +1.3\% for Total-Text and CTW1500).

Aoran Xiao, Jiaxing Huang, Dayan Guan et al.

LiDAR point clouds, which are usually scanned by rotating LiDAR sensors continuously, capture precise geometry of the surrounding environment and are crucial to many autonomous detection and navigation tasks. Though many 3D deep architectures have been developed, efficient collection and annotation of large amounts of point clouds remain one major challenge in the analytic and understanding of point cloud data. This paper presents PolarMix, a point cloud augmentation technique that is simple and generic but can mitigate the data constraint effectively across different perception tasks and scenarios. PolarMix enriches point cloud distributions and preserves point cloud fidelity via two cross-scan augmentation strategies that cut, edit, and mix point clouds along the scanning direction. The first is scene-level swapping which exchanges point cloud sectors of two LiDAR scans that are cut along the azimuth axis. The second is instance-level rotation and paste which crops point instances from one LiDAR scan, rotates them by multiple angles (to create multiple copies), and paste the rotated point instances into other scans. Extensive experiments show that PolarMix achieves superior performance consistently across different perception tasks and scenarios. In addition, it can work as plug-and-play for various 3D deep architectures and also performs well for unsupervised domain adaptation.

Muyu Xu, Fangneng Zhan, Jiahui Zhang et al.

Neural Radiance Field (NeRF) has shown impressive performance in novel view synthesis via implicit scene representation. However, it usually suffers from poor scalability as requiring densely sampled images for each new scene. Several studies have attempted to mitigate this problem by integrating Multi-View Stereo (MVS) technique into NeRF while they still entail a cumbersome fine-tuning process for new scenes. Notably, the rendering quality will drop severely without this fine-tuning process and the errors mainly appear around the high-frequency features. In the light of this observation, we design WaveNeRF, which integrates wavelet frequency decomposition into MVS and NeRF to achieve generalizable yet high-quality synthesis without any per-scene optimization. To preserve high-frequency information when generating 3D feature volumes, WaveNeRF builds Multi-View Stereo in the Wavelet domain by integrating the discrete wavelet transform into the classical cascade MVS, which disentangles high-frequency information explicitly. With that, disentangled frequency features can be injected into classic NeRF via a novel hybrid neural renderer to yield faithful high-frequency details, and an intuitive frequency-guided sampling strategy can be designed to suppress artifacts around high-frequency regions. Extensive experiments over three widely studied benchmarks show that WaveNeRF achieves superior generalizable radiance field modeling when only given three images as input.

Yi Yu, Yufei Wang, Wenhan Yang et al.

Recent deep-learning-based compression methods have achieved superior performance compared with traditional approaches. However, deep learning models have proven to be vulnerable to backdoor attacks, where some specific trigger patterns added to the input can lead to malicious behavior of the models. In this paper, we present a novel backdoor attack with multiple triggers against learned image compression models. Motivated by the widely used discrete cosine transform (DCT) in existing compression systems and standards, we propose a frequency-based trigger injection model that adds triggers in the DCT domain. In particular, we design several attack objectives for various attacking scenarios, including: 1) attacking compression quality in terms of bit-rate and reconstruction quality; 2) attacking task-driven measures, such as down-stream face recognition and semantic segmentation. Moreover, a novel simple dynamic loss is designed to balance the influence of different loss terms adaptively, which helps achieve more efficient training. Extensive experiments show that with our trained trigger injection models and simple modification of encoder parameters (of the compression model), the proposed attack can successfully inject several backdoors with corresponding triggers in a single image compression model.

Gongjie Zhang, Jiahao Lin, Shuang Wu et al.

Vectorized high-definition (HD) map is essential for autonomous driving, providing detailed and precise environmental information for advanced perception and planning. However, current map vectorization methods often exhibit deviations, and the existing evaluation metric for map vectorization lacks sufficient sensitivity to detect these deviations. To address these limitations, we propose integrating the philosophy of rasterization into map vectorization. Specifically, we introduce a new rasterization-based evaluation metric, which has superior sensitivity and is better suited to real-world autonomous driving scenarios. Furthermore, we propose MapVR (Map Vectorization via Rasterization), a novel framework that applies differentiable rasterization to vectorized outputs and then performs precise and geometry-aware supervision on rasterized HD maps. Notably, MapVR designs tailored rasterization strategies for various geometric shapes, enabling effective adaptation to a wide range of map elements. Experiments show that incorporating rasterization into map vectorization greatly enhances performance with no extra computational cost during inference, leading to more accurate map perception and ultimately promoting safer autonomous driving.

Dayan Guan, Jiaxing Huang, Aoran Xiao et al.

Semi-supervised semantic segmentation learns from small amounts of labelled images and large amounts of unlabelled images, which has witnessed impressive progress with the recent advance of deep neural networks. However, it often suffers from severe class-bias problem while exploring the unlabelled images, largely due to the clear pixel-wise class imbalance in the labelled images. This paper presents an unbiased subclass regularization network (USRN) that alleviates the class imbalance issue by learning class-unbiased segmentation from balanced subclass distributions. We build the balanced subclass distributions by clustering pixels of each original class into multiple subclasses of similar sizes, which provide class-balanced pseudo supervision to regularize the class-biased segmentation. In addition, we design an entropy-based gate mechanism to coordinate learning between the original classes and the clustered subclasses which facilitates subclass regularization effectively by suppressing unconfident subclass predictions. Extensive experiments over multiple public benchmarks show that USRN achieves superior performance as compared with the state-of-the-art.

Yingchen Yu, Fangneng Zhan, Rongliang Wu et al.

Leveraging StyleGAN's expressivity and its disentangled latent codes, existing methods can achieve realistic editing of different visual attributes such as age and gender of facial images. An intriguing yet challenging problem arises: Can generative models achieve counterfactual editing against their learnt priors? Due to the lack of counterfactual samples in natural datasets, we investigate this problem in a text-driven manner with Contrastive-Language-Image-Pretraining (CLIP), which can offer rich semantic knowledge even for various counterfactual concepts. Different from in-domain manipulation, counterfactual manipulation requires more comprehensive exploitation of semantic knowledge encapsulated in CLIP as well as more delicate handling of editing directions for avoiding being stuck in local minimum or undesired editing. To this end, we design a novel contrastive loss that exploits predefined CLIP-space directions to guide the editing toward desired directions from different perspectives. In addition, we design a simple yet effective scheme that explicitly maps CLIP embeddings (of target text) to the latent space and fuses them with latent codes for effective latent code optimization and accurate editing. Extensive experiments show that our design achieves accurate and realistic editing while driving by target texts with various counterfactual concepts.

Gongjie Zhang, Zhipeng Luo, Zichen Tian et al.

Multi-scale features have been proven highly effective for object detection but often come with huge and even prohibitive extra computation costs, especially for the recent Transformer-based detectors. In this paper, we propose Iterative Multi-scale Feature Aggregation (IMFA) -- a generic paradigm that enables efficient use of multi-scale features in Transformer-based object detectors. The core idea is to exploit sparse multi-scale features from just a few crucial locations, and it is achieved with two novel designs. First, IMFA rearranges the Transformer encoder-decoder pipeline so that the encoded features can be iteratively updated based on the detection predictions. Second, IMFA sparsely samples scale-adaptive features for refined detection from just a few keypoint locations under the guidance of prior detection predictions. As a result, the sampled multi-scale features are sparse yet still highly beneficial for object detection. Extensive experiments show that the proposed IMFA boosts the performance of multiple Transformer-based object detectors significantly yet with only slight computational overhead.

Chuhui Xue, Zichen Tian, Fangneng Zhan et al.

State-of-the-art document dewarping techniques learn to predict 3-dimensional information of documents which are prone to errors while dealing with documents with irregular distortions or large variations in depth. This paper presents FDRNet, a Fourier Document Restoration Network that can restore documents with different distortions and improve document recognition in a reliable and simpler manner. FDRNet focuses on high-frequency components in the Fourier space that capture most structural information but are largely free of degradation in appearance. It dewarps documents by a flexible Thin-Plate Spline transformation which can handle various deformations effectively without requiring deformation annotations in training. These features allow FDRNet to learn from a small amount of simply labeled training images, and the learned model can dewarp documents with complex geometric distortion and recognize the restored texts accurately. To facilitate document restoration research, we create a benchmark dataset consisting of over one thousand camera documents with different types of geometric and photometric distortion. Extensive experiments show that FDRNet outperforms the state-of-the-art by large margins on both dewarping and text recognition tasks. In addition, FDRNet requires a small amount of simply labeled training data and is easy to deploy.

Zhipeng Luo, Changqing Zhou, Gongjie Zhang et al.

3D object detection with surround-view images is an essential task for autonomous driving. In this work, we propose DETR4D, a Transformer-based framework that explores sparse attention and direct feature query for 3D object detection in multi-view images. We design a novel projective cross-attention mechanism for query-image interaction to address the limitations of existing methods in terms of geometric cue exploitation and information loss for cross-view objects. In addition, we introduce a heatmap generation technique that bridges 3D and 2D spaces efficiently via query initialization. Furthermore, unlike the common practice of fusing intermediate spatial features for temporal aggregation, we provide a new perspective by introducing a novel hybrid approach that performs cross-frame fusion over past object queries and image features, enabling efficient and robust modeling of temporal information. Extensive experiments on the nuScenes dataset demonstrate the effectiveness and efficiency of the proposed DETR4D.

Fangneng Zhan, Yingchen Yu, Rongliang Wu et al.

Exemplar-based image translation establishes dense correspondences between a conditional input and an exemplar (from two different domains) for leveraging detailed exemplar styles to achieve realistic image translation. Existing work builds the cross-domain correspondences implicitly by minimizing feature-wise distances across the two domains. Without explicit exploitation of domain-invariant features, this approach may not reduce the domain gap effectively which often leads to sub-optimal correspondences and image translation. We design a Marginal Contrastive Learning Network (MCL-Net) that explores contrastive learning to learn domain-invariant features for realistic exemplar-based image translation. Specifically, we design an innovative marginal contrastive loss that guides to establish dense correspondences explicitly. Nevertheless, building correspondence with domain-invariant semantics alone may impair the texture patterns and lead to degraded texture generation. We thus design a Self-Correlation Map (SCM) that incorporates scene structures as auxiliary information which improves the built correspondences substantially. Quantitative and qualitative experiments on multifarious image translation tasks show that the proposed method outperforms the state-of-the-art consistently.

Sicong Leng, Hang Zhang, Guanzheng Chen et al.

Large Vision-Language Models (LVLMs) have advanced considerably, intertwining visual recognition and language understanding to generate content that is not only coherent but also contextually attuned. Despite their success, LVLMs still suffer from the issue of object hallucinations, where models generate plausible yet incorrect outputs that include objects that do not exist in the images. To mitigate this issue, we introduce Visual Contrastive Decoding (VCD), a simple and training-free method that contrasts output distributions derived from original and distorted visual inputs. The proposed VCD effectively reduces the over-reliance on statistical bias and unimodal priors, two essential causes of object hallucinations. This adjustment ensures the generated content is closely grounded to visual inputs, resulting in contextually accurate outputs. Our experiments show that VCD, without either additional training or the usage of external tools, significantly mitigates the object hallucination issue across different LVLM families. Beyond mitigating object hallucinations, VCD also excels in general LVLM benchmarks, highlighting its wide-ranging applicability.

Siyuan Yang, Jun Liu, Shijian Lu et al.

3D Skeleton-based human action recognition has attracted increasing attention in recent years. Most of the existing work focuses on supervised learning which requires a large number of labeled action sequences that are often expensive and time-consuming to annotate. In this paper, we address self-supervised 3D action representation learning for skeleton-based action recognition. We investigate self-supervised representation learning and design a novel skeleton cloud colorization technique that is capable of learning spatial and temporal skeleton representations from unlabeled skeleton sequence data. We represent a skeleton action sequence as a 3D skeleton cloud and colorize each point in the cloud according to its temporal and spatial orders in the original (unannotated) skeleton sequence. Leveraging the colorized skeleton point cloud, we design an auto-encoder framework that can learn spatial-temporal features from the artificial color labels of skeleton joints effectively. Specifically, we design a two-steam pretraining network that leverages fine-grained and coarse-grained colorization to learn multi-scale spatial-temporal features. In addition, we design a Masked Skeleton Cloud Repainting task that can pretrain the designed auto-encoder framework to learn informative representations. We evaluate our skeleton cloud colorization approach with linear classifiers trained under different configurations, including unsupervised, semi-supervised, fully-supervised, and transfer learning settings. Extensive experiments on NTU RGB+D, NTU RGB+D 120, PKU-MMD, NW-UCLA, and UWA3D datasets show that the proposed method outperforms existing unsupervised and semi-supervised 3D action recognition methods by large margins and achieves competitive performance in supervised 3D action recognition as well.

Zhipeng Luo, Changqing Zhou, Liang Pan et al.

With the prevalence of LiDAR sensors in autonomous driving, 3D object tracking has received increasing attention. In a point cloud sequence, 3D object tracking aims to predict the location and orientation of an object in consecutive frames given an object template. Motivated by the success of transformers, we propose Point Tracking TRansformer (PTTR), which efficiently predicts high-quality 3D tracking results in a coarse-to-fine manner with the help of transformer operations. PTTR consists of three novel designs. 1) Instead of random sampling, we design Relation-Aware Sampling to preserve relevant points to the given template during subsampling. 2) We propose a Point Relation Transformer for effective feature aggregation and feature matching between the template and search region. 3) Based on the coarse tracking results, we employ a novel Prediction Refinement Module to obtain the final refined prediction through local feature pooling. In addition, motivated by the favorable properties of the Bird's-Eye View (BEV) of point clouds in capturing object motion, we further design a more advanced framework named PTTR++, which incorporates both the point-wise view and BEV representation to exploit their complementary effect in generating high-quality tracking results. PTTR++ substantially boosts the tracking performance on top of PTTR with low computational overhead. Extensive experiments over multiple datasets show that our proposed approaches achieve superior 3D tracking accuracy and efficiency.

Jiahui Zhang, Fangneng Zhan, Rongliang Wu et al.

Neural Radiance Fields (NeRF) have demonstrated very impressive performance in novel view synthesis via implicitly modelling 3D representations from multi-view 2D images. However, most existing studies train NeRF models with either reasonable camera pose initialization or manually-crafted camera pose distributions which are often unavailable or hard to acquire in various real-world data. We design VMRF, an innovative view matching NeRF that enables effective NeRF training without requiring prior knowledge in camera poses or camera pose distributions. VMRF introduces a view matching scheme, which exploits unbalanced optimal transport to produce a feature transport plan for mapping a rendered image with randomly initialized camera pose to the corresponding real image. With the feature transport plan as the guidance, a novel pose calibration technique is designed which rectifies the initially randomized camera poses by predicting relative pose transformations between the pair of rendered and real images. Extensive experiments over a number of synthetic and real datasets show that the proposed VMRF outperforms the state-of-the-art qualitatively and quantitatively by large margins.

Yun Xing, Dayan Guan, Jiaxing Huang et al.

Video semantic segmentation has achieved great progress under the supervision of large amounts of labelled training data. However, domain adaptive video segmentation, which can mitigate data labelling constraints by adapting from a labelled source domain toward an unlabelled target domain, is largely neglected. We design temporal pseudo supervision (TPS), a simple and effective method that explores the idea of consistency training for learning effective representations from unlabelled target videos. Unlike traditional consistency training that builds consistency in spatial space, we explore consistency training in spatiotemporal space by enforcing model consistency across augmented video frames which helps learn from more diverse target data. Specifically, we design cross-frame pseudo labelling to provide pseudo supervision from previous video frames while learning from the augmented current video frames. The cross-frame pseudo labelling encourages the network to produce high-certainty predictions, which facilitates consistency training with cross-frame augmentation effectively. Extensive experiments over multiple public datasets show that TPS is simpler to implement, much more stable to train, and achieves superior video segmentation accuracy as compared with the state-of-the-art.

Siyuan Yang, Jun Liu, Shijian Lu et al.

One-shot skeleton action recognition, which aims to learn a skeleton action recognition model with a single training sample, has attracted increasing interest due to the challenge of collecting and annotating large-scale skeleton action data. However, most existing studies match skeleton sequences by comparing their feature vectors directly which neglects spatial structures and temporal orders of skeleton data. This paper presents a novel one-shot skeleton action recognition technique that handles skeleton action recognition via multi-scale spatial-temporal feature matching. We represent skeleton data at multiple spatial and temporal scales and achieve optimal feature matching from two perspectives. The first is multi-scale matching which captures the scale-wise semantic relevance of skeleton data at multiple spatial and temporal scales simultaneously. The second is cross-scale matching which handles different motion magnitudes and speeds by capturing sample-wise relevance across multiple scales. Extensive experiments over three large-scale datasets (NTU RGB+D, NTU RGB+D 120, and PKU-MMD) show that our method achieves superior one-shot skeleton action recognition, and it outperforms the state-of-the-art consistently by large margins.

Kaiwen Cui, Rongliang Wu, Fangneng Zhan et al.

Face swapping aims to generate swapped images that fuse the identity of source faces and the attributes of target faces. Most existing works address this challenging task through 3D modelling or generation using generative adversarial networks (GANs), but 3D modelling suffers from limited reconstruction accuracy and GANs often struggle in preserving subtle yet important identity details of source faces (e.g., skin colors, face features) and structural attributes of target faces (e.g., face shapes, facial expressions). This paper presents Face Transformer, a novel face swapping network that can accurately preserve source identities and target attributes simultaneously in the swapped face images. We introduce a transformer network for the face swapping task, which learns high-quality semantic-aware correspondence between source and target faces and maps identity features of source faces to the corresponding region in target faces. The high-quality semantic-aware correspondence enables smooth and accurate transfer of source identity information with minimal modification of target shapes and expressions. In addition, our Face Transformer incorporates a multi-scale transformation mechanism for preserving the rich fine facial details. Extensive experiments show that our Face Transformer achieves superior face swapping performance qualitatively and quantitatively.

Fangneng Zhan, Yingchen Yu, Rongliang Wu et al.

Deep generative models have achieved conspicuous progress in realistic image synthesis with multifarious conditional inputs, while generating diverse yet high-fidelity images remains a grand challenge in conditional image generation. This paper presents a versatile framework for conditional image generation which incorporates the inductive bias of CNNs and powerful sequence modeling of auto-regression that naturally leads to diverse image generation. Instead of independently quantizing the features of multiple domains as in prior research, we design an integrated quantization scheme with a variational regularizer that mingles the feature discretization in multiple domains, and markedly boosts the auto-regressive modeling performance. Notably, the variational regularizer enables to regularize feature distributions in incomparable latent spaces by penalizing the intra-domain variations of distributions. In addition, we design a Gumbel sampling strategy that allows to incorporate distribution uncertainty into the auto-regressive training procedure. The Gumbel sampling substantially mitigates the exposure bias that often incurs misalignment between the training and inference stages and severely impairs the inference performance. Extensive experiments over multiple conditional image generation tasks show that our method achieves superior diverse image generation performance qualitatively and quantitatively as compared with the state-of-the-art.

Jiahui Zhang, Fangneng Zhan, Yingchen Yu et al.

Pose-free neural radiance fields (NeRF) aim to train NeRF with unposed multi-view images and it has achieved very impressive success in recent years. Most existing works share the pipeline of training a coarse pose estimator with rendered images at first, followed by a joint optimization of estimated poses and neural radiance field. However, as the pose estimator is trained with only rendered images, the pose estimation is usually biased or inaccurate for real images due to the domain gap between real images and rendered images, leading to poor robustness for the pose estimation of real images and further local minima in joint optimization. We design IR-NeRF, an innovative pose-free NeRF that introduces implicit pose regularization to refine pose estimator with unposed real images and improve the robustness of the pose estimation for real images. With a collection of 2D images of a specific scene, IR-NeRF constructs a scene codebook that stores scene features and captures the scene-specific pose distribution implicitly as priors. Thus, the robustness of pose estimation can be promoted with the scene priors according to the rationale that a 2D real image can be well reconstructed from the scene codebook only when its estimated pose lies within the pose distribution. Extensive experiments show that IR-NeRF achieves superior novel view synthesis and outperforms the state-of-the-art consistently across multiple synthetic and real datasets.

Rongliang Wu, Yingchen Yu, Fangneng Zhan et al.

Audio-driven talking face generation, which aims to synthesize talking faces with realistic facial animations (including accurate lip movements, vivid facial expression details and natural head poses) corresponding to the audio, has achieved rapid progress in recent years. However, most existing work focuses on generating lip movements only without handling the closely correlated facial expressions, which degrades the realism of the generated faces greatly. This paper presents DIRFA, a novel method that can generate talking faces with diverse yet realistic facial animations from the same driving audio. To accommodate fair variation of plausible facial animations for the same audio, we design a transformer-based probabilistic mapping network that can model the variational facial animation distribution conditioned upon the input audio and autoregressively convert the audio signals into a facial animation sequence. In addition, we introduce a temporally-biased mask into the mapping network, which allows to model the temporal dependency of facial animations and produce temporally smooth facial animation sequence. With the generated facial animation sequence and a source image, photo-realistic talking faces can be synthesized with a generic generation network. Extensive experiments show that DIRFA can generate talking faces with realistic facial animations effectively.

Zhipeng Luo, Gongjie Zhang, Changqing Zhou et al.

3D object detection using point clouds has attracted increasing attention due to its wide applications in autonomous driving and robotics. However, most existing studies focus on single point cloud frames without harnessing the temporal information in point cloud sequences. In this paper, we design TransPillars, a novel transformer-based feature aggregation technique that exploits temporal features of consecutive point cloud frames for multi-frame 3D object detection. TransPillars aggregates spatial-temporal point cloud features from two perspectives. First, it fuses voxel-level features directly from multi-frame feature maps instead of pooled instance features to preserve instance details with contextual information that are essential to accurate object localization. Second, it introduces a hierarchical coarse-to-fine strategy to fuse multi-scale features progressively to effectively capture the motion of moving objects and guide the aggregation of fine features. Besides, a variant of deformable transformer is introduced to improve the effectiveness of cross-frame feature matching. Extensive experiments show that our proposed TransPillars achieves state-of-art performance as compared to existing multi-frame detection approaches. Code will be released.

Chuhui Xue, Jiaxing Huang, Shijian Lu et al.

Most existing scene text detectors focus on detecting characters or words that only capture partial text messages due to missing contextual information. For a better understanding of text in scenes, it is more desired to detect contextual text blocks (CTBs) which consist of one or multiple integral text units (e.g., characters, words, or phrases) in natural reading order and transmit certain complete text messages. This paper presents contextual text detection, a new setup that detects CTBs for better understanding of texts in scenes. We formulate the new setup by a dual detection task which first detects integral text units and then groups them into a CTB. To this end, we design a novel scene text clustering technique that treats integral text units as tokens and groups them (belonging to the same CTB) into an ordered token sequence. In addition, we create two datasets SCUT-CTW-Context and ReCTS-Context to facilitate future research, where each CTB is well annotated by an ordered sequence of integral text units. Further, we introduce three metrics that measure contextual text detection in local accuracy, continuity, and global accuracy. Extensive experiments show that our method accurately detects CTBs which effectively facilitates downstream tasks such as text classification and translation. The project is available at https://sg-vilab.github.io/publication/xue2022contextual/.

Jingyi Zhang, Jiaxing Huang, Xiaoqin Zhang et al.

Domain adaptive panoptic segmentation aims to mitigate data annotation challenge by leveraging off-the-shelf annotated data in one or multiple related source domains. However, existing studies employ two separate networks for instance segmentation and semantic segmentation which lead to excessive network parameters as well as complicated and computationally intensive training and inference processes. We design UniDAformer, a unified domain adaptive panoptic segmentation transformer that is simple but can achieve domain adaptive instance segmentation and semantic segmentation simultaneously within a single network. UniDAformer introduces Hierarchical Mask Calibration (HMC) that rectifies inaccurate predictions at the level of regions, superpixels and pixels via online self-training on the fly. It has three unique features: 1) it enables unified domain adaptive panoptic adaptation; 2) it mitigates false predictions and improves domain adaptive panoptic segmentation effectively; 3) it is end-to-end trainable with a much simpler training and inference pipeline. Extensive experiments over multiple public benchmarks show that UniDAformer achieves superior domain adaptive panoptic segmentation as compared with the state-of-the-art.

Zhipeng Luo, Gongjie Zhang, Changqing Zhou et al.

The task of 3D single object tracking (SOT) with LiDAR point clouds is crucial for various applications, such as autonomous driving and robotics. However, existing approaches have primarily relied on appearance matching or motion modeling within only two successive frames, thereby overlooking the long-range continuous motion property of objects in 3D space. To address this issue, this paper presents a novel approach that views each tracklet as a continuous stream: at each timestamp, only the current frame is fed into the network to interact with multi-frame historical features stored in a memory bank, enabling efficient exploitation of sequential information. To achieve effective cross-frame message passing, a hybrid attention mechanism is designed to account for both long-range relation modeling and local geometric feature extraction. Furthermore, to enhance the utilization of multi-frame features for robust tracking, a contrastive sequence enhancement strategy is proposed, which uses ground truth tracklets to augment training sequences and promote discrimination against false positives in a contrastive manner. Extensive experiments demonstrate that the proposed method outperforms the state-of-the-art method by significant margins on multiple benchmarks.

Rongliang Wu, Yingchen Yu, Fangneng Zhan et al.

Facial expression editing has attracted increasing attention with the advance of deep neural networks in recent years. However, most existing methods suffer from compromised editing fidelity and limited usability as they either ignore pose variations (unrealistic editing) or require paired training data (not easy to collect) for pose controls. This paper presents POCE, an innovative pose-controllable expression editing network that can generate realistic facial expressions and head poses simultaneously with just unpaired training images. POCE achieves the more accessible and realistic pose-controllable expression editing by mapping face images into UV space, where facial expressions and head poses can be disentangled and edited separately. POCE has two novel designs. The first is self-supervised UV completion that allows to complete UV maps sampled under different head poses, which often suffer from self-occlusions and missing facial texture. The second is weakly-supervised UV editing that allows to generate new facial expressions with minimal modification of facial identity, where the synthesized expression could be controlled by either an expression label or directly transplanted from a reference UV map via feature transfer. Extensive experiments show that POCE can learn from unpaired face images effectively, and the learned model can generate realistic and high-fidelity facial expressions under various new poses.

Zichen Tian, Chuhui Xue, Jingyi Zhang et al.

Most existing scene text detectors require large-scale training data which cannot scale well due to two major factors: 1) scene text images often have domain-specific distributions; 2) collecting large-scale annotated scene text images is laborious. We study domain adaptive scene text detection, a largely neglected yet very meaningful task that aims for optimal transfer of labelled scene text images while handling unlabelled images in various new domains. Specifically, we design SCAST, a subcategory-aware self-training technique that mitigates the network overfitting and noisy pseudo labels in domain adaptive scene text detection effectively. SCAST consists of two novel designs. For labelled source data, it introduces pseudo subcategories for both foreground texts and background stuff which helps train more generalizable source models with multi-class detection objectives. For unlabelled target data, it mitigates the network overfitting by co-regularizing the binary and subcategory classifiers trained in the source domain. Extensive experiments show that SCAST achieves superior detection performance consistently across multiple public benchmarks, and it also generalizes well to other domain adaptive detection tasks such as vehicle detection.

Jiahui Zhang, Fangneng Zhan, Yingchen Yu et al.

Recently, single image super-resolution (SR) under large scaling factors has witnessed impressive progress by introducing pre-trained generative adversarial networks (GANs) as priors. However, most GAN-Priors based SR methods are constrained by an attribute disentanglement problem in inverted latent codes which directly leads to mismatches of visual attributes in the generator layers and further degraded reconstruction. In addition, stochastic noises fed to the generator are employed for unconditional detail generation, which tends to produce unfaithful details that compromise the fidelity of the generated SR image. We design LAREN, a LAtent multi-Relation rEasoNing technique that achieves superb large-factor SR through graph-based multi-relation reasoning in latent space. LAREN consists of two innovative designs. The first is graph-based disentanglement that constructs a superior disentangled latent space via hierarchical multi-relation reasoning. The second is graph-based code generation that produces image-specific codes progressively via recursive relation reasoning which enables prior GANs to generate desirable image details. Extensive experiments show that LAREN achieves superior large-factor image SR and outperforms the state-of-the-art consistently across multiple benchmarks.

Wenhao Li, Xueying Jiang, Gongjie Zhang et al.

4D point cloud videos capture rich spatial and temporal dynamics of scenes which possess unique values in various 4D understanding tasks. However, most existing methods work in the spatiotemporal domain where the underlying geometric characteristics of 4D point cloud videos are hard to capture, leading to degraded representation learning and understanding of 4D point cloud videos. We address the above challenge from a complementary spectral perspective. By transforming 4D point cloud videos into graph spectral signals, we can decompose them into multiple frequency bands each of which captures distinct geometric structures of point cloud videos. Our spectral analysis reveals that the decomposed low-frequency signals capture more coarse shapes while high-frequency signals encode more fine-grained geometry details. Building on these observations, we design Spatio-Temporal-Spectral Mixer (STS-Mixer), a unified framework that mixes spatial, temporal, and spectral representations of point cloud videos. STS-Mixer integrates multi-band delineated spectral signals with spatiotemporal information to capture rich geometries and temporal dynamics, while enabling fine-grained and holistic understanding of 4D point cloud videos. Extensive experiments show that STS-Mixer achieves superior performance consistently across multiple widely adopted benchmarks on both 3D action recognition and 4D semantic segmentation tasks. Code and models are available at https://github.com/Vegetebird/STS-Mixer.

Yun Xing, Jian Kang, Aoran Xiao et al.

Vision-Language Pre-training has demonstrated its remarkable zero-shot recognition ability and potential to learn generalizable visual representations from language supervision. Taking a step ahead, language-supervised semantic segmentation enables spatial localization of textual inputs by learning pixel grouping solely from image-text pairs. Nevertheless, the state-of-the-art suffers from clear semantic gaps between visual and textual modality: plenty of visual concepts appeared in images are missing in their paired captions. Such semantic misalignment circulates in pre-training, leading to inferior zero-shot performance in dense predictions due to insufficient visual concepts captured in textual representations. To close such semantic gap, we propose Concept Curation (CoCu), a pipeline that leverages CLIP to compensate for the missing semantics. For each image-text pair, we establish a concept archive that maintains potential visually-matched concepts with our proposed vision-driven expansion and text-to-vision-guided ranking. Relevant concepts can thus be identified via cluster-guided sampling and fed into pre-training, thereby bridging the gap between visual and textual semantics. Extensive experiments over a broad suite of 8 segmentation benchmarks show that CoCu achieves superb zero-shot transfer performance and greatly boosts language-supervised segmentation baseline by a large margin, suggesting the value of bridging semantic gap in pre-training data.

Jingyi Zhang, Jiaxing Huang, Xueying Jiang et al.

Black-box unsupervised domain adaptation (UDA) learns with source predictions of target data without accessing either source data or source models during training, and it has clear superiority in data privacy and flexibility in target network selection. However, the source predictions of target data are often noisy and training with them is prone to learning collapses. We propose BiMem, a bi-directional memorization mechanism that learns to remember useful and representative information to correct noisy pseudo labels on the fly, leading to robust black-box UDA that can generalize across different visual recognition tasks. BiMem constructs three types of memory, including sensory memory, short-term memory, and long-term memory, which interact in a bi-directional manner for comprehensive and robust memorization of learnt features. It includes a forward memorization flow that identifies and stores useful features and a backward calibration flow that rectifies features' pseudo labels progressively. Extensive experiments show that BiMem achieves superior domain adaptation performance consistently across various visual recognition tasks such as image classification, semantic segmentation and object detection.

Aoran Xiao, Weihao Xuan, Heli Qi et al.

Robust and accurate segmentation of scenes has become one core functionality in various visual recognition and navigation tasks. This has inspired the recent development of Segment Anything Model (SAM), a foundation model for general mask segmentation. However, SAM is largely tailored for single-modal RGB images, limiting its applicability to multi-modal data captured with widely-adopted sensor suites, such as LiDAR plus RGB, depth plus RGB, thermal plus RGB, etc. We develop MM-SAM, an extension and expansion of SAM that supports cross-modal and multi-modal processing for robust and enhanced segmentation with different sensor suites. MM-SAM features two key designs, namely, unsupervised cross-modal transfer and weakly-supervised multi-modal fusion, enabling label-efficient and parameter-efficient adaptation toward various sensor modalities. It addresses three main challenges: 1) adaptation toward diverse non-RGB sensors for single-modal processing, 2) synergistic processing of multi-modal data via sensor fusion, and 3) mask-free training for different downstream tasks. Extensive experiments show that MM-SAM consistently outperforms SAM by large margins, demonstrating its effectiveness and robustness across various sensors and data modalities.

Jiaxing Huang, Jingyi Zhang, Han Qiu et al.

Traditional domain adaptation assumes the same vocabulary across source and target domains, which often struggles with limited transfer flexibility and efficiency while handling target domains with different vocabularies. Inspired by recent vision-language models (VLMs) that enable open-vocabulary visual recognition by reasoning on both images and texts, we study open-vocabulary domain adaptation (OVDA), a new unsupervised domain adaptation framework that positions a pre-trained VLM as the source model and transfers it towards arbitrary unlabelled target domains. To this end, we design a Prompt Ensemble Self-training (PEST) technique that exploits the synergy between vision and language to mitigate the domain discrepancies in image and text distributions simultaneously. Specifically, PEST makes use of the complementary property of multiple prompts within and across vision and language modalities, which enables joint exploitation of vision and language information and effective learning of image-text correspondences in the unlabelled target domains. Additionally, PEST captures temporal information via temporal prompt ensemble which helps memorize previously learnt target information. Extensive experiments show that PEST outperforms the state-of-the-art consistently across 10 image recognition tasks.

Yi Yu, Qichen Zheng, Siyuan Yang et al.

Deep neural networks are proven to be vulnerable to data poisoning attacks. Recently, a specific type of data poisoning attack known as availability attacks has led to the failure of data utilization for model learning by adding imperceptible perturbations to images. Consequently, it is quite beneficial and challenging to detect poisoned samples, also known as Unlearnable Examples (UEs), from a mixed dataset. In response, we propose an Iterative Filtering approach for UEs identification. This method leverages the distinction between the inherent semantic mapping rules and shortcuts, without the need for any additional information. We verify that when training a classifier on a mixed dataset containing both UEs and clean data, the model tends to quickly adapt to the UEs compared to the clean data. Due to the accuracy gaps between training with clean/poisoned samples, we employ a model to misclassify clean samples while correctly identifying the poisoned ones. The incorporation of additional classes and iterative refinement enhances the model's ability to differentiate between clean and poisoned samples. Extensive experiments demonstrate the superiority of our method over state-of-the-art detection approaches across various attacks, datasets, and poison ratios, significantly reducing the Half Total Error Rate (HTER) compared to existing methods.

Muyu Xu, Fangneng Zhan, Xiaoqin Zhang et al.

Sparse-view 3D Gaussian splatting seeks to render high-quality novel views of 3D scenes from a limited set of input images. While recent pose-free feed-forward methods leveraging pre-trained 3D priors have achieved impressive results, most of them rely on full fine-tuning of large Vision Transformer (ViT) backbones and incur substantial GPU costs. In this work, we introduce MuSASplat, a novel framework that dramatically reduces the computational burden of training pose-free feed-forward 3D Gaussian splats models with little compromise of rendering quality. Central to our approach is a lightweight Multi-Scale Adapter that enables efficient fine-tuning of ViT-based architectures with only a small fraction of training parameters. This design avoids the prohibitive GPU overhead associated with previous full-model adaptation techniques while maintaining high fidelity in novel view synthesis, even with very sparse input views. In addition, we introduce a Feature Fusion Aggregator that integrates features across input views effectively and efficiently. Unlike widely adopted memory banks, the Feature Fusion Aggregator ensures consistent geometric integration across input views and meanwhile mitigates the memory usage, training complexity, and computational costs significantly. Extensive experiments across diverse datasets show that MuSASplat achieves state-of-the-art rendering quality but has significantly reduced parameters and training resource requirements as compared with existing methods.

Aoran Xiao, Weihao Xuan, Junjue Wang et al.

Remote Sensing (RS) is a crucial technology for observing, monitoring, and interpreting our planet, with broad applications across geoscience, economics, humanitarian fields, etc. While artificial intelligence (AI), particularly deep learning, has achieved significant advances in RS, unique challenges persist in developing more intelligent RS systems, including the complexity of Earth's environments, diverse sensor modalities, distinctive feature patterns, varying spatial and spectral resolutions, and temporal dynamics. Meanwhile, recent breakthroughs in large Foundation Models (FMs) have expanded AI's potential across many domains due to their exceptional generalizability and zero-shot transfer capabilities. However, their success has largely been confined to natural data like images and video, with degraded performance and even failures for RS data of various non-optical modalities. This has inspired growing interest in developing Remote Sensing Foundation Models (RSFMs) to address the complex demands of Earth Observation (EO) tasks, spanning the surface, atmosphere, and oceans. This survey systematically reviews the emerging field of RSFMs. It begins with an outline of their motivation and background, followed by an introduction of their foundational concepts. It then categorizes and reviews existing RSFM studies including their datasets and technical contributions across Visual Foundation Models (VFMs), Visual-Language Models (VLMs), Large Language Models (LLMs), and beyond. In addition, we benchmark these models against publicly available datasets, discuss existing challenges, and propose future research directions in this rapidly evolving field. A project associated with this survey has been built at https://github.com/xiaoaoran/awesome-RSFMs .

Yun Xing, Hanyuan Liu, Jiahao Nie et al.

Large Multimodal Models (LMMs) have recently demonstrated their proficiency in holistic visual comprehension. However, most of them struggle to tackle region-level perception guided by visual prompts, especially for cases where multiple regions are referred simultaneously, or scenarios where global contexts are necessary for precise visual referring. We introduce Contextual Latent Steering (CSteer), a training-free approach for guiding general LMMs to refer multiple regions contextually, without expensive fine-tuning or architectural modifications. CSteer starts with pre-computing contextual vectors that implicitly represent visual referring behaviors, such as differentiation among regions and attention to global contexts, followed by representation editing during inference time. Experimental results on multiple datasets indicate that general LMMs with CSteer outperform tailored referring LMMs in most cases, suggesting a promising solution in training-free, and setting new state-of-the-art for this field. Code is available at https://github.com/xing0047/csteer.git.

Yi Yu, Yufei Wang, Song Xia et al.

Unlearnable examples (UEs) seek to maximize testing error by making subtle modifications to training examples that are correctly labeled. Defenses against these poisoning attacks can be categorized based on whether specific interventions are adopted during training. The first approach is training-time defense, such as adversarial training, which can mitigate poisoning effects but is computationally intensive. The other approach is pre-training purification, e.g., image short squeezing, which consists of several simple compressions but often encounters challenges in dealing with various UEs. Our work provides a novel disentanglement mechanism to build an efficient pre-training purification method. Firstly, we uncover rate-constrained variational autoencoders (VAEs), demonstrating a clear tendency to suppress the perturbations in UEs. We subsequently conduct a theoretical analysis for this phenomenon. Building upon these insights, we introduce a disentangle variational autoencoder (D-VAE), capable of disentangling the perturbations with learnable class-wise embeddings. Based on this network, a two-stage purification approach is naturally developed. The first stage focuses on roughly eliminating perturbations, while the second stage produces refined, poison-free results, ensuring effectiveness and robustness across various scenarios. Extensive experiments demonstrate the remarkable performance of our method across CIFAR-10, CIFAR-100, and a 100-class ImageNet-subset. Code is available at https://github.com/yuyi-sd/D-VAE.

Jiahao Nie, Wenbin An, Gongjie Zhang et al.

Despite recent advances in Video Large Language Models (Vid-LLMs), Temporal Video Grounding (TVG), which aims to precisely localize time segments corresponding to query events, remains a significant challenge. Existing methods often match start and end frames by comparing frame features with two separate tokens, relying heavily on exact timestamps. However, this approach fails to capture the event's semantic continuity and integrity, leading to ambiguities. To address this, we propose E.M.Ground, a novel Vid-LLM for TVG that focuses on holistic and coherent event perception. E.M.Ground introduces three key innovations: (i) a special <event> token that aggregates information from all frames of a query event, preserving semantic continuity for accurate event matching; (ii) Savitzky-Golay smoothing to reduce noise in token-to-frame similarities across timestamps, improving prediction accuracy; (iii) multi-grained frame feature aggregation to enhance matching reliability and temporal understanding, compensating for compression-induced information loss. Extensive experiments on benchmark datasets show that E.M.Ground consistently outperforms state-of-the-art Vid-LLMs by significant margins.

Xueying Jiang, Wenhao Li, Quanhao Qian et al.

3D localization in Multimodal Large Language Models (MLLMs), including 3D object detection and 3D visual grounding, is fundamentally limited by camera intrinsic ambiguity: the same image admits different 3D scenes under different cameras. Existing MLLMs either ignore camera parameters and overfit to a canonical training intrinsic, or retrieve depth and 3D cues from external tools but treat the returned values as reference cues (numerical hints that the model is free to interpret implicitly), both preventing camera information from being deterministically propagated into the prediction. We propose an equation-anchored tool-use framework that re-purposes spatial tools as formula variables. The proposed framework proactively retrieves camera intrinsics and samples multi-point metric depths, writes the pinhole back-projection equation $\hat{X} = (u_c - c_x)\bar{Z}/f_x$ explicitly in Chain-of-Thought (CoT), and substitutes tool outputs into the formula before regressing the final 9-DoF bounding box. On both 3D object detection and 3D visual grounding tasks under rescaled camera intrinsics from $0.5\times$ to $1.5\times$, our method outperforms RGB-only and tool-augmented baselines, with significant gains where the camera deviates most from the training scale. Code and data will be released.

Zuhao Yang, Kaichen Zhang, Sudong Wang et al.

Training large multimodal models (LMMs) via reinforcement learning (RL) to natively invoke video-processing tools (e.g., cropping) has become a promising route to long-video understanding. However, existing native-RL methods dispatch tool calls sequentially (i.e., one per turn): a single wrong crop propagates errors without peer correction, multi-turn tool calls corrupt context, and inference cost scales linearly with the number of turns. We introduce ParaVT, the first multi-agent end-to-end RL-trained framework for Parallel Video Tool calling, dispatching multiple time-window crops in a single turn for cleaner context and better fault tolerance. Yet applying standard RL to ParaVT reveals an obstacle we term the Tool Prior Paradox: the pretrained tool priors that enable tool exploration also destabilize cold-started structural format and expose the skip-tool reward shortcut under temperature sampling. A cross-model contrast on a weaker-prior LMM supports this claim: format stays stable but RL elicits zero tool calls, indicating that prior strength is the shared driver of both format collapse and tool exploration. We propose PARA-GRPO (Parseability-Anchored and Ratio-gAted GRPO), which augments standard RL with two complementary mechanisms: (i) a targeted format reward applied only at the structural-token positions most prone to collapse, and (ii) a per-prompt frame-budget randomization that creates training prompts where calling the tool yields a measurable reward signal over skipping it. Across six long-video understanding benchmarks, ParaVT improves over the Qwen3-VL baseline by +7.9% on average, with PARA-GRPO lifting training-time format compliance from 0.13 to 0.64. As tool capabilities become increasingly internalized in modern LMMs, RL must cooperate with the resulting priors, and ParaVT offers a general recipe for agentic RL. Code, data, and model weights are publicly available.