Yu‐Wing Tai

Benran Hu, Junkai Huang, Yichen Liu et al. · cmu

This paper presents the first significant object detection framework, NeRF-RPN, which directly operates on NeRF. Given a pre-trained NeRF model, NeRF-RPN aims to detect all bounding boxes of objects in a scene. By exploiting a novel voxel representation that incorporates multi-scale 3D neural volumetric features, we demonstrate it is possible to regress the 3D bounding boxes of objects in NeRF directly without rendering the NeRF at any viewpoint. NeRF-RPN is a general framework and can be applied to detect objects without class labels. We experimented NeRF-RPN with various backbone architectures, RPN head designs and loss functions. All of them can be trained in an end-to-end manner to estimate high quality 3D bounding boxes. To facilitate future research in object detection for NeRF, we built a new benchmark dataset which consists of both synthetic and real-world data with careful labeling and clean up. Code and dataset are available at https://github.com/lyclyc52/NeRF_RPN.

Yichen Liu, Benran Hu, Junkai Huang et al. · cmu

This paper presents one of the first learning-based NeRF 3D instance segmentation pipelines, dubbed as Instance Neural Radiance Field, or Instance NeRF. Taking a NeRF pretrained from multi-view RGB images as input, Instance NeRF can learn 3D instance segmentation of a given scene, represented as an instance field component of the NeRF model. To this end, we adopt a 3D proposal-based mask prediction network on the sampled volumetric features from NeRF, which generates discrete 3D instance masks. The coarse 3D mask prediction is then projected to image space to match 2D segmentation masks from different views generated by existing panoptic segmentation models, which are used to supervise the training of the instance field. Notably, beyond generating consistent 2D segmentation maps from novel views, Instance NeRF can query instance information at any 3D point, which greatly enhances NeRF object segmentation and manipulation. Our method is also one of the first to achieve such results in pure inference. Experimented on synthetic and real-world NeRF datasets with complex indoor scenes, Instance NeRF surpasses previous NeRF segmentation works and competitive 2D segmentation methods in segmentation performance on unseen views. Watch the demo video at https://youtu.be/wW9Bme73coI. Code and data are available at https://github.com/lyclyc52/Instance_NeRF.

Lei Ke, Mingqiao Ye, Martin Danelljan et al.

The recent Segment Anything Model (SAM) represents a big leap in scaling up segmentation models, allowing for powerful zero-shot capabilities and flexible prompting. Despite being trained with 1.1 billion masks, SAM's mask prediction quality falls short in many cases, particularly when dealing with objects that have intricate structures. We propose HQ-SAM, equipping SAM with the ability to accurately segment any object, while maintaining SAM's original promptable design, efficiency, and zero-shot generalizability. Our careful design reuses and preserves the pre-trained model weights of SAM, while only introducing minimal additional parameters and computation. We design a learnable High-Quality Output Token, which is injected into SAM's mask decoder and is responsible for predicting the high-quality mask. Instead of only applying it on mask-decoder features, we first fuse them with early and final ViT features for improved mask details. To train our introduced learnable parameters, we compose a dataset of 44K fine-grained masks from several sources. HQ-SAM is only trained on the introduced detaset of 44k masks, which takes only 4 hours on 8 GPUs. We show the efficacy of HQ-SAM in a suite of 10 diverse segmentation datasets across different downstream tasks, where 8 out of them are evaluated in a zero-shot transfer protocol. Our code and pretrained models are at https://github.com/SysCV/SAM-HQ.

Qi Fan, Wenjie Pei, Yu-Wing Tai et al.

Existing few-shot segmentation methods have achieved great progress based on the support-query matching framework. But they still heavily suffer from the limited coverage of intra-class variations from the few-shot supports provided. Motivated by the simple Gestalt principle that pixels belonging to the same object are more similar than those to different objects of same class, we propose a novel self-support matching strategy to alleviate this problem, which uses query prototypes to match query features, where the query prototypes are collected from high-confidence query predictions. This strategy can effectively capture the consistent underlying characteristics of the query objects, and thus fittingly match query features. We also propose an adaptive self-support background prototype generation module and self-support loss to further facilitate the self-support matching procedure. Our self-support network substantially improves the prototype quality, benefits more improvement from stronger backbones and more supports, and achieves SOTA on multiple datasets. Codes are at \url{https://github.com/fanq15/SSP}.

Frano Rajič, Lei Ke, Yu-Wing Tai et al.

The Segment Anything Model (SAM) has established itself as a powerful zero-shot image segmentation model, enabled by efficient point-centric annotation and prompt-based models. While click and brush interactions are both well explored in interactive image segmentation, the existing methods on videos focus on mask annotation and propagation. This paper presents SAM-PT, a novel method for point-centric interactive video segmentation, empowered by SAM and long-term point tracking. SAM-PT leverages robust and sparse point selection and propagation techniques for mask generation. Compared to traditional object-centric mask propagation strategies, we uniquely use point propagation to exploit local structure information agnostic to object semantics. We highlight the merits of point-based tracking through direct evaluation on the zero-shot open-world Unidentified Video Objects (UVO) benchmark. Our experiments on popular video object segmentation and multi-object segmentation tracking benchmarks, including DAVIS, YouTube-VOS, and BDD100K, suggest that a point-based segmentation tracker yields better zero-shot performance and efficient interactions. We release our code that integrates different point trackers and video segmentation benchmarks at https://github.com/SysCV/sam-pt.

Xinpeng Liu, Yong-Lu Li, Xiaoqian Wu et al.

Human-Object Interaction (HOI) detection plays a core role in activity understanding. Though recent two/one-stage methods have achieved impressive results, as an essential step, discovering interactive human-object pairs remains challenging. Both one/two-stage methods fail to effectively extract interactive pairs instead of generating redundant negative pairs. In this work, we introduce a previously overlooked interactiveness bimodal prior: given an object in an image, after pairing it with the humans, the generated pairs are either mostly non-interactive, or mostly interactive, with the former more frequent than the latter. Based on this interactiveness bimodal prior we propose the "interactiveness field". To make the learned field compatible with real HOI image considerations, we propose new energy constraints based on the cardinality and difference in the inherent "interactiveness field" underlying interactive versus non-interactive pairs. Consequently, our method can detect more precise pairs and thus significantly boost HOI detection performance, which is validated on widely-used benchmarks where we achieve decent improvements over state-of-the-arts. Our code is available at https://github.com/Foruck/Interactiveness-Field.

Mingqiao Ye, Lei Ke, Siyuan Li et al.

Object localization in general environments is a fundamental part of vision systems. While dominating on the COCO benchmark, recent Transformer-based detection methods are not competitive in diverse domains. Moreover, these methods still struggle to very accurately estimate the object bounding boxes in complex environments. We introduce Cascade-DETR for high-quality universal object detection. We jointly tackle the generalization to diverse domains and localization accuracy by proposing the Cascade Attention layer, which explicitly integrates object-centric information into the detection decoder by limiting the attention to the previous box prediction. To further enhance accuracy, we also revisit the scoring of queries. Instead of relying on classification scores, we predict the expected IoU of the query, leading to substantially more well-calibrated confidences. Lastly, we introduce a universal object detection benchmark, UDB10, that contains 10 datasets from diverse domains. While also advancing the state-of-the-art on COCO, Cascade-DETR substantially improves DETR-based detectors on all datasets in UDB10, even by over 10 mAP in some cases. The improvements under stringent quality requirements are even more pronounced. Our code and models will be released at https://github.com/SysCV/cascade-detr.

Yanan Sun, Chi-Keung Tang, Yu-Wing Tai

This paper introduces a new matting task called human instance matting (HIM), which requires the pertinent model to automatically predict a precise alpha matte for each human instance. Straightforward combination of closely related techniques, namely, instance segmentation, soft segmentation and human/conventional matting, will easily fail in complex cases requiring disentangling mingled colors belonging to multiple instances along hairy and thin boundary structures. To tackle these technical challenges, we propose a human instance matting framework, called InstMatt, where a novel mutual guidance strategy working in tandem with a multi-instance refinement module is used, for delineating multi-instance relationship among humans with complex and overlapping boundaries if present. A new instance matting metric called instance matting quality (IMQ) is proposed, which addresses the absence of a unified and fair means of evaluation emphasizing both instance recognition and matting quality. Finally, we construct a HIM benchmark for evaluation, which comprises of both synthetic and natural benchmark images. In addition to thorough experimental results on complex cases with multiple and overlapping human instances each has intricate boundaries, preliminary results are presented on general instance matting. Code and benchmark are available in https://github.com/nowsyn/InstMatt.

Lei Ke, Martin Danelljan, Henghui Ding et al.

The recent advancement in Video Instance Segmentation (VIS) has largely been driven by the use of deeper and increasingly data-hungry transformer-based models. However, video masks are tedious and expensive to annotate, limiting the scale and diversity of existing VIS datasets. In this work, we aim to remove the mask-annotation requirement. We propose MaskFreeVIS, achieving highly competitive VIS performance, while only using bounding box annotations for the object state. We leverage the rich temporal mask consistency constraints in videos by introducing the Temporal KNN-patch Loss (TK-Loss), providing strong mask supervision without any labels. Our TK-Loss finds one-to-many matches across frames, through an efficient patch-matching step followed by a K-nearest neighbor selection. A consistency loss is then enforced on the found matches. Our mask-free objective is simple to implement, has no trainable parameters, is computationally efficient, yet outperforms baselines employing, e.g., state-of-the-art optical flow to enforce temporal mask consistency. We validate MaskFreeVIS on the YouTube-VIS 2019/2021, OVIS and BDD100K MOTS benchmarks. The results clearly demonstrate the efficacy of our method by drastically narrowing the gap between fully and weakly-supervised VIS performance. Our code and trained models are available at https://github.com/SysCV/MaskFreeVis.

Lei Ke, Yu-Wing Tai, Chi-Keung Tang

Segmenting highly-overlapping image objects is challenging, because there is typically no distinction between real object contours and occlusion boundaries on images. Unlike previous instance segmentation methods, we model image formation as a composition of two overlapping layers, and propose Bilayer Convolutional Network (BCNet), where the top layer detects occluding objects (occluders) and the bottom layer infers partially occluded instances (occludees). The explicit modeling of occlusion relationship with bilayer structure naturally decouples the boundaries of both the occluding and occluded instances, and considers the interaction between them during mask regression. We investigate the efficacy of bilayer structure using two popular convolutional network designs, namely, Fully Convolutional Network (FCN) and Graph Convolutional Network (GCN). Further, we formulate bilayer decoupling using the vision transformer (ViT), by representing instances in the image as separate learnable occluder and occludee queries. Large and consistent improvements using one/two-stage and query-based object detectors with various backbones and network layer choices validate the generalization ability of bilayer decoupling, as shown by extensive experiments on image instance segmentation benchmarks (COCO, KINS, COCOA) and video instance segmentation benchmarks (YTVIS, OVIS, BDD100K MOTS), especially for heavy occlusion cases. Code and data are available at https://github.com/lkeab/BCNet.

Hao Zhang, Tianyuan Dai, Yu-Wing Tai et al.

This paper presents the first significant work on directly predicting 3D face landmarks on neural radiance fields (NeRFs). Our 3D coarse-to-fine Face Landmarks NeRF (FLNeRF) model efficiently samples from a given face NeRF with individual facial features for accurate landmarks detection. Expression augmentation is applied to facial features in a fine scale to simulate large emotions range including exaggerated facial expressions (e.g., cheek blowing, wide opening mouth, eye blinking) for training FLNeRF. Qualitative and quantitative comparison with related state-of-the-art 3D facial landmark estimation methods demonstrate the efficacy of FLNeRF, which contributes to downstream tasks such as high-quality face editing and swapping with direct control using our NeRF landmarks. Code and data will be available. Github link: https://github.com/ZHANG1023/FLNeRF.

Takashi Isobe, Xu Jia, Xin Tao et al.

Temporal modeling is crucial for video super-resolution. Most of the video super-resolution methods adopt the optical flow or deformable convolution for explicitly motion compensation. However, such temporal modeling techniques increase the model complexity and might fail in case of occlusion or complex motion, resulting in serious distortion and artifacts. In this paper, we propose to explore the role of explicit temporal difference modeling in both LR and HR space. Instead of directly feeding consecutive frames into a VSR model, we propose to compute the temporal difference between frames and divide those pixels into two subsets according to the level of difference. They are separately processed with two branches of different receptive fields in order to better extract complementary information. To further enhance the super-resolution result, not only spatial residual features are extracted, but the difference between consecutive frames in high-frequency domain is also computed. It allows the model to exploit intermediate SR results in both future and past to refine the current SR output. The difference at different time steps could be cached such that information from further distance in time could be propagated to the current frame for refinement. Experiments on several video super-resolution benchmark datasets demonstrate the effectiveness of the proposed method and its favorable performance against state-of-the-art methods.

Qi Fan, Xin Tao, Lei Ke et al.

The Segment Anything Model (SAM) achieves remarkable promptable segmentation given high-quality prompts which, however, often require good skills to specify. To make SAM robust to casual prompts, this paper presents the first comprehensive analysis on SAM's segmentation stability across a diverse spectrum of prompt qualities, notably imprecise bounding boxes and insufficient points. Our key finding reveals that given such low-quality prompts, SAM's mask decoder tends to activate image features that are biased towards the background or confined to specific object parts. To mitigate this issue, our key idea consists of calibrating solely SAM's mask attention by adjusting the sampling locations and amplitudes of image features, while the original SAM model architecture and weights remain unchanged. Consequently, our deformable sampling plugin (DSP) enables SAM to adaptively shift attention to the prompted target regions in a data-driven manner, facilitated by our effective robust training strategy (RTS). During inference, dynamic routing plugin (DRP) is proposed that toggles SAM between the deformable and regular grid sampling modes, conditioned on the input prompt quality. Thus, our solution, termed Stable-SAM, offers several advantages: 1) improved SAM's segmentation stability across a wide range of prompt qualities, while 2) retaining SAM's powerful promptable segmentation efficiency and generality, with 3) minimal learnable parameters (0.08 M) and fast adaptation (by 1 training epoch). Extensive experiments across multiple datasets validate the effectiveness and advantages of our approach, underscoring Stable-SAM as a more robust solution for segmenting anything. Codes will be released upon acceptance. https://github.com/fanq15/Stable-SAM

Lei Ke, Henghui Ding, Martin Danelljan et al.

While Video Instance Segmentation (VIS) has seen rapid progress, current approaches struggle to predict high-quality masks with accurate boundary details. Moreover, the predicted segmentations often fluctuate over time, suggesting that temporal consistency cues are neglected or not fully utilized. In this paper, we set out to tackle these issues, with the aim of achieving highly detailed and more temporally stable mask predictions for VIS. We first propose the Video Mask Transfiner (VMT) method, capable of leveraging fine-grained high-resolution features thanks to a highly efficient video transformer structure. Our VMT detects and groups sparse error-prone spatio-temporal regions of each tracklet in the video segment, which are then refined using both local and instance-level cues. Second, we identify that the coarse boundary annotations of the popular YouTube-VIS dataset constitute a major limiting factor. Based on our VMT architecture, we therefore design an automated annotation refinement approach by iterative training and self-correction. To benchmark high-quality mask predictions for VIS, we introduce the HQ-YTVIS dataset, consisting of a manually re-annotated test set and our automatically refined training data. We compare VMT with the most recent state-of-the-art methods on the HQ-YTVIS, as well as the Youtube-VIS, OVIS and BDD100K MOTS benchmarks. Experimental results clearly demonstrate the efficacy and effectiveness of our method on segmenting complex and dynamic objects, by capturing precise details.

Qi Fan, Mattia Segu, Yu-Wing Tai et al.

Improving model's generalizability against domain shifts is crucial, especially for safety-critical applications such as autonomous driving. Real-world domain styles can vary substantially due to environment changes and sensor noises, but deep models only know the training domain style. Such domain style gap impedes model generalization on diverse real-world domains. Our proposed Normalization Perturbation (NP) can effectively overcome this domain style overfitting problem. We observe that this problem is mainly caused by the biased distribution of low-level features learned in shallow CNN layers. Thus, we propose to perturb the channel statistics of source domain features to synthesize various latent styles, so that the trained deep model can perceive diverse potential domains and generalizes well even without observations of target domain data in training. We further explore the style-sensitive channels for effective style synthesis. Normalization Perturbation only relies on a single source domain and is surprisingly effective and extremely easy to implement. Extensive experiments verify the effectiveness of our method for generalizing models under real-world domain shifts.

Xinhang Liu, Jiaben Chen, Huai Yu et al.

We present radiance field propagation (RFP), a novel approach to segmenting objects in 3D during reconstruction given only unlabeled multi-view images of a scene. RFP is derived from emerging neural radiance field-based techniques, which jointly encodes semantics with appearance and geometry. The core of our method is a novel propagation strategy for individual objects' radiance fields with a bidirectional photometric loss, enabling an unsupervised partitioning of a scene into salient or meaningful regions corresponding to different object instances. To better handle complex scenes with multiple objects and occlusions, we further propose an iterative expectation-maximization algorithm to refine object masks. RFP is one of the first unsupervised approach for tackling 3D real scene object segmentation for neural radiance field (NeRF) without any supervision, annotations, or other cues such as 3D bounding boxes and prior knowledge of object class. Experiments demonstrate that RFP achieves feasible segmentation results that are more accurate than previous unsupervised image/scene segmentation approaches, and are comparable to existing supervised NeRF-based methods. The segmented object representations enable individual 3D object editing operations.

Jiayin Cai, Changlin Li, Xin Tao et al.

This paper proposes a novel video inpainting method. We make three main contributions: First, we extended previous Transformers with patch alignment by introducing Deformed Patch-based Homography (DePtH), which improves patch-level feature alignments without additional supervision and benefits challenging scenes with various deformation. Second, we introduce Mask Pruning-based Patch Attention (MPPA) to improve patch-wised feature matching by pruning out less essential features and using saliency map. MPPA enhances matching accuracy between warped tokens with invalid pixels. Third, we introduce a Spatial-Temporal weighting Adaptor (STA) module to obtain accurate attention to spatial-temporal tokens under the guidance of the Deformation Factor learned from DePtH, especially for videos with agile motions. Experimental results demonstrate that our method outperforms recent methods qualitatively and quantitatively and achieves a new state-of-the-art.

Songlin Tang, Wenjie Pei, Xin Tao et al.

Existing methods for interactive segmentation in radiance fields entail scene-specific optimization and thus cannot generalize across different scenes, which greatly limits their applicability. In this work we make the first attempt at Scene-Generalizable Interactive Segmentation in Radiance Fields (SGISRF) and propose a novel SGISRF method, which can perform 3D object segmentation for novel (unseen) scenes represented by radiance fields, guided by only a few interactive user clicks in a given set of multi-view 2D images. In particular, the proposed SGISRF focuses on addressing three crucial challenges with three specially designed techniques. First, we devise the Cross-Dimension Guidance Propagation to encode the scarce 2D user clicks into informative 3D guidance representations. Second, the Uncertainty-Eliminated 3D Segmentation module is designed to achieve efficient yet effective 3D segmentation. Third, Concealment-Revealed Supervised Learning scheme is proposed to reveal and correct the concealed 3D segmentation errors resulted from the supervision in 2D space with only 2D mask annotations. Extensive experiments on two real-world challenging benchmarks covering diverse scenes demonstrate 1) effectiveness and scene-generalizability of the proposed method, 2) favorable performance compared to classical method requiring scene-specific optimization.

Rui Qin, Bin Wang, Yu-Wing Tai

Text image super-resolution is a unique and important task to enhance readability of text images to humans. It is widely used as pre-processing in scene text recognition. However, due to the complex degradation in natural scenes, recovering high-resolution texts from the low-resolution inputs is ambiguous and challenging. Existing methods mainly leverage deep neural networks trained with pixel-wise losses designed for natural image reconstruction, which ignore the unique character characteristics of texts. A few works proposed content-based losses. However, they only focus on text recognizers' accuracy, while the reconstructed images may still be ambiguous to humans. Further, they often have weak generalizability to handle cross languages. To this end, we present TATSR, a Text-Aware Text Super-Resolution framework, which effectively learns the unique text characteristics using Criss-Cross Transformer Blocks (CCTBs) and a novel Content Perceptual (CP) Loss. The CCTB extracts vertical and horizontal content information from text images by two orthogonal transformers, respectively. The CP Loss supervises the text reconstruction with content semantics by multi-scale text recognition features, which effectively incorporates content awareness into the framework. Extensive experiments on various language datasets demonstrate that TATSR outperforms state-of-the-art methods in terms of both recognition accuracy and human perception.

Shengnan Liang, Yichen Liu, Shangzhe Wu et al.

We present ONeRF, a method that automatically segments and reconstructs object instances in 3D from multi-view RGB images without any additional manual annotations. The segmented 3D objects are represented using separate Neural Radiance Fields (NeRFs) which allow for various 3D scene editing and novel view rendering. At the core of our method is an unsupervised approach using the iterative Expectation-Maximization algorithm, which effectively aggregates 2D visual features and the corresponding 3D cues from multi-views for joint 3D object segmentation and reconstruction. Unlike existing approaches that can only handle simple objects, our method produces segmented full 3D NeRFs of individual objects with complex shapes, topologies and appearance. The segmented ONeRfs enable a range of 3D scene editing, such as object transformation, insertion and deletion.

Xinhang Liu, Yu-Wing Tai, Chi-Keung Tang

While Neural Radiance Fields (NeRFs) had achieved unprecedented novel view synthesis results, they have been struggling in dealing with large-scale cluttered scenes with sparse input views and highly view-dependent appearances. Specifically, existing NeRF-based models tend to produce blurry rendering with the volumetric reconstruction often inaccurate, where a lot of reconstruction errors are observed in the form of foggy "floaters" hovering within the entire volume of an opaque 3D scene. Such inaccuracies impede NeRF's potential for accurate 3D NeRF registration, object detection, segmentation, etc., which possibly accounts for only limited significant research effort so far to directly address these important 3D fundamental computer vision problems to date. This paper analyzes the NeRF's struggles in such settings and proposes Clean-NeRF for accurate 3D reconstruction and novel view rendering in complex scenes. Our key insights consist of enforcing effective appearance and geometry constraints, which are absent in the conventional NeRF reconstruction, by 1) automatically detecting and modeling view-dependent appearances in the training views to prevent them from interfering with density estimation, which is complete with 2) a geometric correction procedure performed on each traced ray during inference. Clean-NeRF can be implemented as a plug-in that can immediately benefit existing NeRF-based methods without additional input. Codes will be released.

Huimin Zeng, Weinong Wang, Xin Tao et al.

Recent interactive segmentation methods iteratively take source image, user guidance and previously predicted mask as the input without considering the invariant nature of the source image. As a result, extracting features from the source image is repeated in each interaction, resulting in substantial computational redundancy. In this work, we propose the Feature Decoupling-Recycling Network (FDRN), which decouples the modeling components based on their intrinsic discrepancies and then recycles components for each user interaction. Thus, the efficiency of the whole interactive process can be significantly improved. To be specific, we apply the Decoupling-Recycling strategy from three perspectives to address three types of discrepancies, respectively. First, our model decouples the learning of source image semantics from the encoding of user guidance to process two types of input domains separately. Second, FDRN decouples high-level and low-level features from stratified semantic representations to enhance feature learning. Third, during the encoding of user guidance, current user guidance is decoupled from historical guidance to highlight the effect of current user guidance. We conduct extensive experiments on 6 datasets from different domains and modalities, which demonstrate the following merits of our model: 1) superior efficiency than other methods, particularly advantageous in challenging scenarios requiring long-term interactions (up to 4.25x faster), while achieving favorable segmentation performance; 2) strong applicability to various methods serving as a universal enhancement technique; 3) well cross-task generalizability, e.g., to medical image segmentation, and robustness against misleading user guidance.

Changlin Li, Guangyang Wu, Yanan Sun et al.

Capitalizing on the rapid development of neural networks, recent video frame interpolation (VFI) methods have achieved notable improvements. However, they still fall short for real-world videos containing large motions. Complex deformation and/or occlusion caused by large motions make it an extremely difficult problem in video frame interpolation. In this paper, we propose a simple yet effective solution, H-VFI, to deal with large motions in video frame interpolation. H-VFI contributes a hierarchical video interpolation transformer (HVIT) to learn a deformable kernel in a coarse-to-fine strategy in multiple scales. The learnt deformable kernel is then utilized in convolving the input frames for predicting the interpolated frame. Starting from the smallest scale, H-VFI updates the deformable kernel by a residual in succession based on former predicted kernels, intermediate interpolated results and hierarchical features from transformer. Bias and masks to refine the final outputs are then predicted by a transformer block based on interpolated results. The advantage of such a progressive approximation is that the large motion frame interpolation problem can be decomposed into several relatively simpler sub-tasks, which enables a very accurate prediction in the final results. Another noteworthy contribution of our paper consists of a large-scale high-quality dataset, YouTube200K, which contains videos depicting a great variety of scenarios captured at high resolution and high frame rate. Extensive experiments on multiple frame interpolation benchmarks validate that H-VFI outperforms existing state-of-the-art methods especially for videos with large motions.

Yanan Sun, Zihan Zhong, Qi Fan et al.

Large-scale joint training of multimodal models, e.g., CLIP, have demonstrated great performance in many vision-language tasks. However, image-text pairs for pre-training are restricted to the intersection of images and texts, limiting their ability to cover a large distribution of real-world data, where noise can also be introduced as misaligned pairs during pre-processing. Conversely, unimodal models trained on text or image data alone through unsupervised techniques can achieve broader coverage of diverse real-world data and are not constrained by the requirement of simultaneous presence of image and text. In this paper, we demonstrate that using large-scale unsupervised unimodal models as pre-training can enhance the zero-shot performance of image-text pair models. Our thorough studies validate that models pre-trained as such can learn rich representations of both modalities, improving their ability to understand how images and text relate to each other. Our experiments show that unimodal pre-training outperforms state-of-the-art CLIP-based models by 6.5% (52.3% $\rightarrow$ 58.8%) on PASCAL-5$^i$ and 6.2% (27.2% $\rightarrow$ 33.4%) on COCO-20$^i$ semantic segmentation under zero-shot setting respectively. By learning representations of both modalities, unimodal pre-training offers broader coverage, reduced misalignment errors, and the ability to capture more complex features and patterns in the real-world data resulting in better performance especially for zero-shot vision-language tasks.

Wenjie Pei, Xin Feng, Canmiao Fu et al.

The key challenge of sequence representation learning is to capture the long-range temporal dependencies. Typical methods for supervised sequence representation learning are built upon recurrent neural networks to capture temporal dependencies. One potential limitation of these methods is that they only model one-order information interactions explicitly between adjacent time steps in a sequence, hence the high-order interactions between nonadjacent time steps are not fully exploited. It greatly limits the capability of modeling the long-range temporal dependencies since the temporal features learned by one-order interactions cannot be maintained for a long term due to temporal information dilution and gradient vanishing. To tackle this limitation, we propose the Non-local Recurrent Neural Memory (NRNM) for supervised sequence representation learning, which performs non-local operations \MR{by means of self-attention mechanism} to learn full-order interactions within a sliding temporal memory block and models global interactions between memory blocks in a gated recurrent manner. Consequently, our model is able to capture long-range dependencies. Besides, the latent high-level features contained in high-order interactions can be distilled by our model. We validate the effectiveness and generalization of our NRNM on three types of sequence applications across different modalities, including sequence classification, step-wise sequential prediction and sequence similarity learning. Our model compares favorably against other state-of-the-art methods specifically designed for each of these sequence applications.

Andy Xu, Yu-Wing Tai

AI agents are increasingly used to solve complex, multi-step tasks, but existing multi-agent frameworks remain brittle as workflows grow in scale and depth. Small errors at intermediate stages can propagate through agent interactions, while insufficient grounding and weak verification mechanisms further limit reliability. We present Meta-Agent, a two-phase framework that automatically constructs and executes specialized multi-agent systems from natural-language task descriptions. In the construction phase, a task planner decomposes a problem into a directed acyclic graph of agent specifications with explicit input/output contracts and verification criteria. A web search module grounds each specification with external evidence, and a code generation module produces system prompts and tool configurations. A construction-time verification stage then validates generated artifacts and triggers targeted regeneration when failures are detected. In the execution phase, a coordinator dispatches subtasks across the agent graph while execution-time verification gates intermediate outputs. We further introduce a three-level error attribution mechanism that distinguishes local, upstream, and structural failures, enabling targeted recovery strategies ranging from localized retries to partial re-execution and re-decomposition. We evaluate Meta-Agent across coding, contextual learning, and open-ended reasoning tasks. Experiments against strong multi-agent baselines and ablation studies demonstrate consistent improvements in task success rate, error recovery, and workflow stability. The results highlight the importance of tightly integrating planning, grounding, and verification for building reliable multi-agent systems.

Yue Xu, Yong-Lu Li, Zhemin Huang et al.

With the surge in attention to Egocentric Hand-Object Interaction (Ego-HOI), large-scale datasets such as Ego4D and EPIC-KITCHENS have been proposed. However, most current research is built on resources derived from third-person video action recognition. This inherent domain gap between first- and third-person action videos, which have not been adequately addressed before, makes current Ego-HOI suboptimal. This paper rethinks and proposes a new framework as an infrastructure to advance Ego-HOI recognition by Probing, Curation and Adaption (EgoPCA). We contribute comprehensive pre-train sets, balanced test sets and a new baseline, which are complete with a training-finetuning strategy. With our new framework, we not only achieve state-of-the-art performance on Ego-HOI benchmarks but also build several new and effective mechanisms and settings to advance further research. We believe our data and the findings will pave a new way for Ego-HOI understanding. Code and data are available at https://mvig-rhos.com/ego_pca

Xinhang Liu, Yu-Wing Tai, Chi-Keung Tang

Cinematographers adeptly capture the essence of the world, crafting compelling visual narratives through intricate camera movements. Witnessing the strides made by large language models in perceiving and interacting with the 3D world, this study explores their capability to control cameras with human language guidance. We introduce ChatCam, a system that navigates camera movements through conversations with users, mimicking a professional cinematographer's workflow. To achieve this, we propose CineGPT, a GPT-based autoregressive model for text-conditioned camera trajectory generation. We also develop an Anchor Determinator to ensure precise camera trajectory placement. ChatCam understands user requests and employs our proposed tools to generate trajectories, which can be used to render high-quality video footage on radiance field representations. Our experiments, including comparisons to state-of-the-art approaches and user studies, demonstrate our approach's ability to interpret and execute complex instructions for camera operation, showing promising applications in real-world production settings.

Juntao Zhang, Yuehuai Liu, Yu-Wing Tai et al.

We present Compound Conditioned ControlNet, C3Net, a novel generative neural architecture taking conditions from multiple modalities and synthesizing multimodal contents simultaneously (e.g., image, text, audio). C3Net adapts the ControlNet architecture to jointly train and make inferences on a production-ready diffusion model and its trainable copies. Specifically, C3Net first aligns the conditions from multi-modalities to the same semantic latent space using modality-specific encoders based on contrastive training. Then, it generates multimodal outputs based on the aligned latent space, whose semantic information is combined using a ControlNet-like architecture called Control C3-UNet. Correspondingly, with this system design, our model offers an improved solution for joint-modality generation through learning and explaining multimodal conditions instead of simply taking linear interpolations on the latent space. Meanwhile, as we align conditions to a unified latent space, C3Net only requires one trainable Control C3-UNet to work on multimodal semantic information. Furthermore, our model employs unimodal pretraining on the condition alignment stage, outperforming the non-pretrained alignment even on relatively scarce training data and thus demonstrating high-quality compound condition generation. We contribute the first high-quality tri-modal validation set to validate quantitatively that C3Net outperforms or is on par with first and contemporary state-of-the-art multimodal generation. Our codes and tri-modal dataset will be released.

Yuyao Zhang, Jinghao Li, Yu-Wing Tai

Text-to-image (T2I) generation has made remarkable progress, yet existing systems still lack intuitive control over spatial composition, object consistency, and multi-step editing. We present $\textbf{LayerCraft}$, a modular framework that uses large language models (LLMs) as autonomous agents to orchestrate structured, layered image generation and editing. LayerCraft supports two key capabilities: (1) $\textit{structured generation}$ from simple prompts via chain-of-thought (CoT) reasoning, enabling it to decompose scenes, reason about object placement, and guide composition in a controllable, interpretable manner; and (2) $\textit{layered object integration}$, allowing users to insert and customize objects -- such as characters or props -- across diverse images or scenes while preserving identity, context, and style. The system comprises a coordinator agent, the $\textbf{ChainArchitect}$ for CoT-driven layout planning, and the $\textbf{Object Integration Network (OIN)}$ for seamless image editing using off-the-shelf T2I models without retraining. Through applications like batch collage editing and narrative scene generation, LayerCraft empowers non-experts to iteratively design, customize, and refine visual content with minimal manual effort. Code will be released at https://github.com/PeterYYZhang/LayerCraft.

Shiu-hong Kao, Xinhang Liu, Yu-Wing Tai et al.

This paper presents InceptionHuman, a prompt-to-NeRF framework that allows easy control via a combination of prompts in different modalities (e.g., text, poses, edge, segmentation map, etc) as inputs to generate photorealistic 3D humans. While many works have focused on generating 3D human models, they suffer one or more of the following: lack of distinctive features, unnatural shading/shadows, unnatural poses/clothes, limited views, etc. InceptionHuman achieves consistent 3D human generation within a progressively refined NeRF space with two novel modules, Iterative Pose-Aware Refinement (IPAR) and Progressive-Augmented Reconstruction (PAR). IPAR iteratively refines the diffusion-generated images and synthesizes high-quality 3D-aware views considering the close-pose RGB values. PAR employs a pretrained diffusion prior to augment the generated synthetic views and adds regularization for view-independent appearance. Overall, the synthesis of photorealistic novel views empowers the resulting 3D human NeRF from 360-degree perspectives. Extensive qualitative and quantitative experimental comparison show that our InceptionHuman models achieve state-of-the-art application quality.

Yufan Deng, Ruida Wang, Yuhao Zhang et al.

Video generation models have shown their superior ability to generate photo-realistic video. However, how to accurately control (or edit) the video remains a formidable challenge. The main issues are: 1) how to perform direct and accurate user control in editing; 2) how to execute editings like changing shape, expression, and layout without unsightly distortion and artifacts to the edited content; and 3) how to maintain spatio-temporal consistency of video after editing. To address the above issues, we propose DragVideo, a general drag-style video editing framework. Inspired by DragGAN, DragVideo addresses issues 1) and 2) by proposing the drag-style video latent optimization method which gives desired control by updating noisy video latent according to drag instructions through video-level drag objective function. We amend issue 3) by integrating the video diffusion model with sample-specific LoRA and Mutual Self-Attention in DragVideo to ensure the edited result is spatio-temporally consistent. We also present a series of testing examples for drag-style video editing and conduct extensive experiments across a wide array of challenging editing tasks, such as motion, skeleton editing, etc, underscoring DragVideo can edit video in an intuitive, faithful to the user's intention manner, with nearly unnoticeable distortion and artifacts, while maintaining spatio-temporal consistency. While traditional prompt-based video editing fails to do the former two and directly applying image drag editing fails in the last, DragVideo's versatility and generality are emphasized. Github link: https://github.com/RickySkywalker/DragVideo-Official.

Yue Xu, Yong-Lu Li, Kaitong Cui et al.

Data-efficient learning has garnered significant attention, especially given the current trend of large multi-modal models. Recently, dataset distillation has become an effective approach by synthesizing data samples that are essential for network training. However, it remains to be explored which samples are essential for the dataset distillation process itself. In this work, we study the data efficiency and selection for the dataset distillation task. By re-formulating the dynamics of distillation, we provide insight into the inherent redundancy in the real dataset, both theoretically and empirically. We propose to use the empirical loss value as a static data pruning criterion. To further compensate for the variation of the data value in training, we find the most contributing samples based on their causal effects on the distillation. The proposed selection strategy can efficiently exploit the training dataset, outperform the previous SOTA distillation algorithms, and consistently enhance the distillation algorithms, even on much larger-scale and more heterogeneous datasets, e.g., full ImageNet-1K and Kinetics-400. We believe this paradigm will open up new avenues in the dynamics of distillation and pave the way for efficient dataset distillation. Our code is available on https://github.com/silicx/GoldFromOres-BiLP.

Li Xu, Gang He, Jinjia Zhou et al.

In most video platforms, such as Youtube, and TikTok, the played videos usually have undergone multiple video encodings such as hardware encoding by recording devices, software encoding by video editing apps, and single/multiple video transcoding by video application servers. Previous works in compressed video restoration typically assume the compression artifacts are caused by one-time encoding. Thus, the derived solution usually does not work very well in practice. In this paper, we propose a new method, temporal spatial auxiliary network (TSAN), for transcoded video restoration. Our method considers the unique traits between video encoding and transcoding, and we consider the initial shallow encoded videos as the intermediate labels to assist the network to conduct self-supervised attention training. In addition, we employ adjacent multi-frame information and propose the temporal deformable alignment and pyramidal spatial fusion for transcoded video restoration. The experimental results demonstrate that the performance of the proposed method is superior to that of the previous techniques. The code is available at https://github.com/icecherylXuli/TSAN.

Qi Fan, Chi-Keung Tang, Yu-Wing Tai

We introduce Few-Shot Video Object Detection (FSVOD) with three contributions to real-world visual learning challenge in our highly diverse and dynamic world: 1) a large-scale video dataset FSVOD-500 comprising of 500 classes with class-balanced videos in each category for few-shot learning; 2) a novel Tube Proposal Network (TPN) to generate high-quality video tube proposals for aggregating feature representation for the target video object which can be highly dynamic; 3) a strategically improved Temporal Matching Network (TMN+) for matching representative query tube features with better discriminative ability thus achieving higher diversity. Our TPN and TMN+ are jointly and end-to-end trained. Extensive experiments demonstrate that our method produces significantly better detection results on two few-shot video object detection datasets compared to image-based methods and other naive video-based extensions. Codes and datasets are released at \url{https://github.com/fanq15/FewX}.

Lei Ke, Yu-Wing Tai, Chi-Keung Tang

Segmenting highly-overlapping objects is challenging, because typically no distinction is made between real object contours and occlusion boundaries. Unlike previous two-stage instance segmentation methods, we model image formation as composition of two overlapping layers, and propose Bilayer Convolutional Network (BCNet), where the top GCN layer detects the occluding objects (occluder) and the bottom GCN layer infers partially occluded instance (occludee). The explicit modeling of occlusion relationship with bilayer structure naturally decouples the boundaries of both the occluding and occluded instances, and considers the interaction between them during mask regression. We validate the efficacy of bilayer decoupling on both one-stage and two-stage object detectors with different backbones and network layer choices. Despite its simplicity, extensive experiments on COCO and KINS show that our occlusion-aware BCNet achieves large and consistent performance gain especially for heavy occlusion cases. Code is available at https://github.com/lkeab/BCNet.

Yang You, Yujing Lou, Ruoxi Shi et al.

Point cloud analysis without pose priors is very challenging in real applications, as the orientations of point clouds are often unknown. In this paper, we propose a brand new point-set learning framework PRIN, namely, Point-wise Rotation Invariant Network, focusing on rotation invariant feature extraction in point clouds analysis. We construct spherical signals by Density Aware Adaptive Sampling to deal with distorted point distributions in spherical space. Spherical Voxel Convolution and Point Re-sampling are proposed to extract rotation invariant features for each point. In addition, we extend PRIN to a sparse version called SPRIN, which directly operates on sparse point clouds. Both PRIN and SPRIN can be applied to tasks ranging from object classification, part segmentation, to 3D feature matching and label alignment. Results show that, on the dataset with randomly rotated point clouds, SPRIN demonstrates better performance than state-of-the-art methods without any data augmentation. We also provide thorough theoretical proof and analysis for point-wise rotation invariance achieved by our methods. Our code is available on https://github.com/qq456cvb/SPRIN.

Jianfeng Yan, Zizhuang Wei, Hongwei Yi et al.

In this paper, we propose an efficient and effective dense hybrid recurrent multi-view stereo net with dynamic consistency checking, namely $D^{2}$HC-RMVSNet, for accurate dense point cloud reconstruction. Our novel hybrid recurrent multi-view stereo net consists of two core modules: 1) a light DRENet (Dense Reception Expanded) module to extract dense feature maps of original size with multi-scale context information, 2) a HU-LSTM (Hybrid U-LSTM) to regularize 3D matching volume into predicted depth map, which efficiently aggregates different scale information by coupling LSTM and U-Net architecture. To further improve the accuracy and completeness of reconstructed point clouds, we leverage a dynamic consistency checking strategy instead of prefixed parameters and strategies widely adopted in existing methods for dense point cloud reconstruction. In doing so, we dynamically aggregate geometric consistency matching error among all the views. Our method ranks \textbf{$1^{st}$} on the complex outdoor \textsl{Tanks and Temples} benchmark over all the methods. Extensive experiments on the in-door DTU dataset show our method exhibits competitive performance to the state-of-the-art method while dramatically reduces memory consumption, which costs only $19.4\%$ of R-MVSNet memory consumption. The codebase is available at \hyperlink{https://github.com/yhw-yhw/D2HC-RMVSNet}{https://github.com/yhw-yhw/D2HC-RMVSNet}.

Hongwei Yi, Zizhuang Wei, Mingyu Ding et al.

n this paper, we propose an effective and efficient pyramid multi-view stereo (MVS) net with self-adaptive view aggregation for accurate and complete dense point cloud reconstruction. Different from using mean square variance to generate cost volume in previous deep-learning based MVS methods, our \textbf{VA-MVSNet} incorporates the cost variances in different views with small extra memory consumption by introducing two novel self-adaptive view aggregations: pixel-wise view aggregation and voxel-wise view aggregation. To further boost the robustness and completeness of 3D point cloud reconstruction, we extend VA-MVSNet with pyramid multi-scale images input as \textbf{PVA-MVSNet}, where multi-metric constraints are leveraged to aggregate the reliable depth estimation at the coarser scale to fill in the mismatched regions at the finer scale. Experimental results show that our approach establishes a new state-of-the-art on the \textsl{\textbf{DTU}} dataset with significant improvements in the completeness and overall quality, and has strong generalization by achieving a comparable performance as the state-of-the-art methods on the \textsl{\textbf{Tanks and Temples}} benchmark. Our codebase is at \hyperlink{https://github.com/yhw-yhw/PVAMVSNet}{https://github.com/yhw-yhw/PVAMVSNet}

Qi Fan, Wei Zhuo, Chi-Keung Tang et al.

Conventional methods for object detection typically require a substantial amount of training data and preparing such high-quality training data is very labor-intensive. In this paper, we propose a novel few-shot object detection network that aims at detecting objects of unseen categories with only a few annotated examples. Central to our method are our Attention-RPN, Multi-Relation Detector and Contrastive Training strategy, which exploit the similarity between the few shot support set and query set to detect novel objects while suppressing false detection in the background. To train our network, we contribute a new dataset that contains 1000 categories of various objects with high-quality annotations. To the best of our knowledge, this is one of the first datasets specifically designed for few-shot object detection. Once our few-shot network is trained, it can detect objects of unseen categories without further training or fine-tuning. Our method is general and has a wide range of potential applications. We produce a new state-of-the-art performance on different datasets in the few-shot setting. The dataset link is https://github.com/fanq15/Few-Shot-Object-Detection-Dataset.

Xiang Li, Tianhan Wei, Yau Pun Chen et al.

Over the past few years, we have witnessed the success of deep learning in image recognition thanks to the availability of large-scale human-annotated datasets such as PASCAL VOC, ImageNet, and COCO. Although these datasets have covered a wide range of object categories, there are still a significant number of objects that are not included. Can we perform the same task without a lot of human annotations? In this paper, we are interested in few-shot object segmentation where the number of annotated training examples are limited to 5 only. To evaluate and validate the performance of our approach, we have built a few-shot segmentation dataset, FSS-1000, which consists of 1000 object classes with pixelwise annotation of ground-truth segmentation. Unique in FSS-1000, our dataset contains significant number of objects that have never been seen or annotated in previous datasets, such as tiny daily objects, merchandise, cartoon characters, logos, etc. We build our baseline model using standard backbone networks such as VGG-16, ResNet-101, and Inception. To our surprise, we found that training our model from scratch using FSS-1000 achieves comparable and even better results than training with weights pre-trained by ImageNet which is more than 100 times larger than FSS-1000. Both our approach and dataset are simple, effective, and easily extensible to learn segmentation of new object classes given very few annotated training examples. Dataset is available at https://github.com/HKUSTCV/FSS-1000.

Hao-Shu Fang, Guansong Lu, Xiaolin Fang et al.

Human body part parsing, or human semantic part segmentation, is fundamental to many computer vision tasks. In conventional semantic segmentation methods, the ground truth segmentations are provided, and fully convolutional networks (FCN) are trained in an end-to-end scheme. Although these methods have demonstrated impressive results, their performance highly depends on the quantity and quality of training data. In this paper, we present a novel method to generate synthetic human part segmentation data using easily-obtained human keypoint annotations. Our key idea is to exploit the anatomical similarity among human to transfer the parsing results of a person to another person with similar pose. Using these estimated results as additional training data, our semi-supervised model outperforms its strong-supervised counterpart by 6 mIOU on the PASCAL-Person-Part dataset, and we achieve state-of-the-art human parsing results. Our approach is general and can be readily extended to other object/animal parsing task assuming that their anatomical similarity can be annotated by keypoints. The proposed model and accompanying source code are available at https://github.com/MVIG-SJTU/WSHP

Han Jiang, Haosen Sun, Ruoxuan Li et al.

Current Neural Radiance Fields (NeRF) can generate photorealistic novel views. For editing 3D scenes represented by NeRF, with the advent of generative models, this paper proposes Inpaint4DNeRF to capitalize on state-of-the-art stable diffusion models (e.g., ControlNet) for direct generation of the underlying completed background content, regardless of static or dynamic. The key advantages of this generative approach for NeRF inpainting are twofold. First, after rough mask propagation, to complete or fill in previously occluded content, we can individually generate a small subset of completed images with plausible content, called seed images, from which simple 3D geometry proxies can be derived. Second and the remaining problem is thus 3D multiview consistency among all completed images, now guided by the seed images and their 3D proxies. Without other bells and whistles, our generative Inpaint4DNeRF baseline framework is general which can be readily extended to 4D dynamic NeRFs, where temporal consistency can be naturally handled in a similar way as our multiview consistency.

Hao Zhang, Yu-Wing Tai, Chi-Keung Tang

The success of the GAN-NeRF structure has enabled face editing on NeRF to maintain 3D view consistency. However, achieving simultaneously multi-view consistency and temporal coherence while editing video sequences remains a formidable challenge. This paper proposes a novel face video editing architecture built upon the dynamic face GAN-NeRF structure, which effectively utilizes video sequences to restore the latent code and 3D face geometry. By editing the latent code, multi-view consistent editing on the face can be ensured, as validated by multiview stereo reconstruction on the resulting edited images in our dynamic NeRF. As the estimation of face geometries occurs on a frame-by-frame basis, this may introduce a jittering issue. We propose a stabilizer that maintains temporal coherence by preserving smooth changes of face expressions in consecutive frames. Quantitative and qualitative analyses reveal that our method, as the pioneering 4D face video editor, achieves state-of-the-art performance in comparison to existing 2D or 3D-based approaches independently addressing identity and motion. Codes will be released.

Alexander Huang-Menders, Xinhang Liu, Andy Xu et al.

SmartAvatar is a vision-language-agent-driven framework for generating fully rigged, animation-ready 3D human avatars from a single photo or textual prompt. While diffusion-based methods have made progress in general 3D object generation, they continue to struggle with precise control over human identity, body shape, and animation readiness. In contrast, SmartAvatar leverages the commonsense reasoning capabilities of large vision-language models (VLMs) in combination with off-the-shelf parametric human generators to deliver high-quality, customizable avatars. A key innovation is an autonomous verification loop, where the agent renders draft avatars, evaluates facial similarity, anatomical plausibility, and prompt alignment, and iteratively adjusts generation parameters for convergence. This interactive, AI-guided refinement process promotes fine-grained control over both facial and body features, enabling users to iteratively refine their avatars via natural-language conversations. Unlike diffusion models that rely on static pre-trained datasets and offer limited flexibility, SmartAvatar brings users into the modeling loop and ensures continuous improvement through an LLM-driven procedural generation and verification system. The generated avatars are fully rigged and support pose manipulation with consistent identity and appearance, making them suitable for downstream animation and interactive applications. Quantitative benchmarks and user studies demonstrate that SmartAvatar outperforms recent text- and image-driven avatar generation systems in terms of reconstructed mesh quality, identity fidelity, attribute accuracy, and animation readiness, making it a versatile tool for realistic, customizable avatar creation on consumer-grade hardware.

Jincheng Yan, Yun Wang, Xiaoyan Luo et al.

Person re-identification (ReID) plays a critical role in applications such as security surveillance and criminal investigations. Most traditional image-based ReID methods face challenges including occlusions and lighting changes, while text provides complementary information to mitigate these issues. However, the integration of both image and text modalities remains underexplored. To address this gap, we propose {\bf PS-ReID}, a multimodal model that combines image and text inputs to enhance ReID performance. In contrast to existing ReID methods limited by cropped pedestrian images, our PS-ReID focuses on full-scene settings and introduces a multimodal ReID task that incorporates segmentation, enabling precise feature extraction of the queried individual, even under challenging conditions such as occlusion. To this end, our model adopts a dual-path asymmetric encoding scheme that explicitly separates query and target roles: the query branch captures identity-discriminative cues, while the target branch performs holistic scene reasoning. Additionally, a token-level ReID loss supervises identity-aware tokens, coupling retrieval and segmentation to yield masks that are both spatially precise and identity-consistent. To facilitate systematic evaluation, we construct M2ReID, currently the largest full-scene multimodal ReID dataset, with over 200K images and 4,894 identities, featuring multimodal queries and high-quality segmentation masks. Experimental results demonstrate that PS-ReID significantly outperforms unimodal query-based models in both ReID and segmentation tasks. The model excels in challenging real-world scenarios such as occlusion, low lighting, and background clutter, offering a robust and flexible solution for person retrieval and segmentation. All code, models, and datasets will be publicly available.

Jianmeng Liu, Yuyao Zhang, Zeyuan Meng et al.

This paper explores promptable NeRF generation (e.g., text prompt or single image prompt) for direct conditioning and fast generation of NeRF parameters for the underlying 3D scenes, thus undoing complex intermediate steps while providing full 3D generation with conditional control. Unlike previous diffusion-CLIP-based pipelines that involve tedious per-prompt optimizations, Prompt2NeRF-PIL is capable of generating a variety of 3D objects with a single forward pass, leveraging a pre-trained implicit latent space of NeRF parameters. Furthermore, in zero-shot tasks, our experiments demonstrate that the NeRFs produced by our method serve as semantically informative initializations, significantly accelerating the inference process of existing prompt-to-NeRF methods. Specifically, we will show that our approach speeds up the text-to-NeRF model DreamFusion and the 3D reconstruction speed of the image-to-NeRF method Zero-1-to-3 by 3 to 5 times.

Junwei Zhou, Yu-Wing Tai

Reconstructing 3D objects from a few unposed and partially occluded views is a common yet challenging problem in real-world scenarios, where many object surfaces are never directly observed. Traditional multi-view or inpainting-based approaches struggle under such conditions, often yielding incomplete or geometrically inconsistent reconstructions. We introduce AmodalGen3D, a generative framework for amodal 3D object reconstruction that infers complete, occlusion-free geometry and appearance from arbitrary sparse inputs. The model integrates 2D amodal completion priors with multi-view stereo geometry conditioning, supported by a View-Wise Cross Attention mechanism for sparse-view feature fusion and a Stereo-Conditioned Cross Attention module for unobserved structure inference. By jointly modeling visible and hidden regions, AmodalGen3D faithfully reconstructs 3D objects that are consistent with sparse-view constraints while plausibly hallucinating unseen parts. Experiments on both synthetic and real-world datasets demonstrate that AmodalGen3D achieves superior fidelity and completeness under occlusion-heavy sparse-view settings, addressing a pressing need for object-level 3D scene reconstruction in robotics, AR/VR, and embodied AI applications.

Yuyao Zhang, Yu-Wing Tai

Ultra-high-resolution text-to-image generation demands both fine-grained texture synthesis and globally coherent structure, yet current diffusion models remain constrained to sub-$1K \times 1K$ resolutions due to the prohibitive quadratic complexity of attention and the scarcity of native $4K$ training data. We present \textbf{Scale-DiT}, a new diffusion framework that introduces hierarchical local attention with low-resolution global guidance, enabling efficient, scalable, and semantically coherent image synthesis at ultra-high resolutions. Specifically, high-resolution latents are divided into fixed-size local windows to reduce attention complexity from quadratic to near-linear, while a low-resolution latent equipped with scaled positional anchors injects global semantics. A lightweight LoRA adaptation bridges global and local pathways during denoising, ensuring consistency across structure and detail. To maximize inference efficiency, we repermute token sequence in Hilbert curve order and implement a fused-kernel for skipping masked operations, resulting in a GPU-friendly design. Extensive experiments demonstrate that Scale-DiT achieves more than $2\times$ faster inference and lower memory usage compared to dense attention baselines, while reliably scaling to $4K \times 4K$ resolution without requiring additional high-resolution training data. On both quantitative benchmarks (FID, IS, CLIP Score) and qualitative comparisons, Scale-DiT delivers superior global coherence and sharper local detail, matching or outperforming state-of-the-art methods that rely on native 4K training. Taken together, these results highlight hierarchical local attention with guided low-resolution anchors as a promising and effective approach for advancing ultra-high-resolution image generation.

Xinhang Liu, Yuxi Xiao, Donny Y. Chen et al. · bytedance

Effective spatio-temporal representation is fundamental to modeling, understanding, and predicting dynamics in videos. The atomic unit of a video, the pixel, traces a continuous 3D trajectory over time, serving as the primitive element of dynamics. Based on this principle, we propose representing any video as a Trajectory Field: a dense mapping that assigns a continuous 3D trajectory function of time to each pixel in every frame. With this representation, we introduce Trace Anything, a neural network that predicts the entire trajectory field in a single feed-forward pass. Specifically, for each pixel in each frame, our model predicts a set of control points that parameterizes a trajectory (i.e., a B-spline), yielding its 3D position at arbitrary query time instants. We trained the Trace Anything model on large-scale 4D data, including data from our new platform, and our experiments demonstrate that: (i) Trace Anything achieves state-of-the-art performance on our new benchmark for trajectory field estimation and performs competitively on established point-tracking benchmarks; (ii) it offers significant efficiency gains thanks to its one-pass paradigm, without requiring iterative optimization or auxiliary estimators; and (iii) it exhibits emergent abilities, including goal-conditioned manipulation, motion forecasting, and spatio-temporal fusion. Project page: https://trace-anything.github.io/.