Search

Embodied Everyday Task is a popular task in the embodied AI community, requiring agents to make a sequence of actions based on natural language instructions and visual observations. Traditional learning-based approaches face two challenges. Firstly, natural language instructions often lack explicit task planning. Secondly, extensive training is required to equip models with knowledge of the task environment. Previous works based on Large Language Model (LLM) either suffer from poor performance due to the lack of task-specific knowledge or rely on ground truth as few-shot samples. To address the above limitations, we propose a novel approach called Progressive Retrieval Augmented Generation (P-RAG), which not only effectively leverages the powerful language processing capabilities of LLMs but also progressively accumulates task-specific knowledge without ground-truth. Compared to the conventional RAG methods, which retrieve relevant information from the database in a one-shot manner to assist generation, P-RAG introduces an iterative approach to progressively update the database. In each iteration, P-RAG retrieves the latest database and obtains historical information from the previous interaction as experiential references for the current interaction. Moreover, we also introduce a more granular retrieval scheme that not only retrieves similar tasks but also incorporates retrieval of similar situations to provide more valuable reference experiences. Extensive experiments reveal that P-RAG achieves competitive results without utilizing ground truth and can even further improve performance through self-iterations.

Large Language Models (LLMs) tend to prioritize adherence to user prompts over providing veracious responses, leading to the sycophancy issue. When challenged by users, LLMs tend to admit mistakes and provide inaccurate responses even if they initially provided the correct answer. Recent works propose to employ supervised fine-tuning (SFT) to mitigate the sycophancy issue, while it typically leads to the degeneration of LLMs' general capability. To address the challenge, we propose a novel supervised pinpoint tuning (SPT), where the region-of-interest modules are tuned for a given objective. Specifically, SPT first reveals and verifies a small percentage (<5%) of the basic modules, which significantly affect a particular behavior of LLMs. i.e., sycophancy. Subsequently, SPT merely fine-tunes these identified modules while freezing the rest. To verify the effectiveness of the proposed SPT, we conduct comprehensive experiments, demonstrating that SPT significantly mitigates the sycophancy issue of LLMs (even better than SFT). Moreover, SPT introduces limited or even no side effects on the general capability of LLMs. Our results shed light on how to precisely, effectively, and efficiently explain and improve the targeted ability of LLMs., Comment: Accepted by ICML 2024

Over the past few years, the advancement of Multimodal Large Language Models (MLLMs) has captured the wide interest of researchers, leading to numerous innovations to enhance MLLMs' comprehension. In this paper, we present AdaptVision, a multimodal large language model specifically designed to dynamically process input images at varying resolutions. We hypothesize that the requisite number of visual tokens for the model is contingent upon both the resolution and content of the input image. Generally, natural images with a lower information density can be effectively interpreted by the model using fewer visual tokens at reduced resolutions. In contrast, images containing textual content, such as documents with rich text, necessitate a higher number of visual tokens for accurate text interpretation due to their higher information density. Building on this insight, we devise a dynamic image partitioning module that adjusts the number of visual tokens according to the size and aspect ratio of images. This method mitigates distortion effects that arise from resizing images to a uniform resolution and dynamically optimizing the visual tokens input to the LLMs. Our model is capable of processing images with resolutions up to $1008\times 1008$. Extensive experiments across various datasets demonstrate that our method achieves impressive performance in handling vision-language tasks in both natural and text-related scenes. The source code and dataset are now publicly available at \url{https://github.com/harrytea/AdaptVision}.

In video lane detection, there are rich temporal contexts among successive frames, which is under-explored in existing lane detectors. In this work, we propose LaneTCA to bridge the individual video frames and explore how to effectively aggregate the temporal context. Technically, we develop an accumulative attention module and an adjacent attention module to abstract the long-term and short-term temporal context, respectively. The accumulative attention module continuously accumulates visual information during the journey of a vehicle, while the adjacent attention module propagates this lane information from the previous frame to the current frame. The two modules are meticulously designed based on the transformer architecture. Finally, these long-short context features are fused with the current frame features to predict the lane lines in the current frame. Extensive quantitative and qualitative experiments are conducted on two prevalent benchmark datasets. The results demonstrate the effectiveness of our method, achieving several new state-of-the-art records. The codes and models are available at https://github.com/Alex-1337/LaneTCA

Sign language serves as the primary meaning of communication for the deaf-mute community. Different from spoken language, it commonly conveys information by the collaboration of manual features, i.e., hand gestures and body movements, and non-manual features, i.e., facial expressions and mouth cues. To facilitate communication between the deaf-mute and hearing people, a series of sign language understanding (SLU) tasks have been studied in recent years, including isolated/continuous sign language recognition (ISLR/CSLR), gloss-free sign language translation (GF-SLT) and sign language retrieval (SL-RT). Sign language recognition and translation aims to understand the semantic meaning conveyed by sign languages from gloss-level and sentence-level, respectively. In contrast, SL-RT focuses on retrieving sign videos or corresponding texts from a closed-set under the query-by-example search paradigm. These tasks investigate sign language topics from diverse perspectives and raise challenges in learning effective representation of sign language videos. To advance the development of sign language understanding, exploring a generalized model that is applicable across various SLU tasks is a profound research direction., Comment: Sign language recognition; Sign language translation; Sign language retrieval

Shadow detection is a fundamental and challenging task in many computer vision applications. Intuitively, most shadows come from the occlusion of light by the object itself, resulting in the object and its shadow being contiguous (referred to as the adjacent shadow in this paper). In this case, when the color of the object is similar to that of the shadow, existing methods struggle to achieve accurate detection. To address this problem, we present SwinShadow, a transformer-based architecture that fully utilizes the powerful shifted window mechanism for detecting adjacent shadows. The mechanism operates in two steps. Initially, it applies local self-attention within a single window, enabling the network to focus on local details. Subsequently, it shifts the attention windows to facilitate inter-window attention, enabling the capture of a broader range of adjacent information. These combined steps significantly improve the network's capacity to distinguish shadows from nearby objects. And the whole process can be divided into three parts: encoder, decoder, and feature integration. During encoding, we adopt Swin Transformer to acquire hierarchical features. Then during decoding, for shallow layers, we propose a deep supervision (DS) module to suppress the false positives and boost the representation capability of shadow features for subsequent processing, while for deep layers, we leverage a double attention (DA) module to integrate local and shifted window in one stage to achieve a larger receptive field and enhance the continuity of information. Ultimately, a new multi-level aggregation (MLA) mechanism is applied to fuse the decoded features for mask prediction. Extensive experiments on three shadow detection benchmark datasets, SBU, UCF, and ISTD, demonstrate that our network achieves good performance in terms of balance error rate (BER).

Different from traditional video retrieval, sign language retrieval is more biased towards understanding the semantic information of human actions contained in video clips. Previous works typically only encode RGB videos to obtain high-level semantic features, resulting in local action details drowned in a large amount of visual information redundancy. Furthermore, existing RGB-based sign retrieval works suffer from the huge memory cost of dense visual data embedding in end-to-end training, and adopt offline RGB encoder instead, leading to suboptimal feature representation. To address these issues, we propose a novel sign language representation framework called Semantically Enhanced Dual-Stream Encoder (SEDS), which integrates Pose and RGB modalities to represent the local and global information of sign language videos. Specifically, the Pose encoder embeds the coordinates of keypoints corresponding to human joints, effectively capturing detailed action features. For better context-aware fusion of two video modalities, we propose a Cross Gloss Attention Fusion (CGAF) module to aggregate the adjacent clip features with similar semantic information from intra-modality and inter-modality. Moreover, a Pose-RGB Fine-grained Matching Objective is developed to enhance the aggregated fusion feature by contextual matching of fine-grained dual-stream features. Besides the offline RGB encoder, the whole framework only contains learnable lightweight networks, which can be trained end-to-end. Extensive experiments demonstrate that our framework significantly outperforms state-of-the-art methods on various datasets., Comment: Accepted to ACM International Conference on Multimedia (MM) 2024

Synthesizing realistic 3D indoor scenes is a challenging task that traditionally relies on manual arrangement and annotation by expert designers. Recent advances in autoregressive models have automated this process, but they often lack semantic understanding of the relationships and hierarchies present in real-world scenes, yielding limited performance. In this paper, we propose Forest2Seq, a framework that formulates indoor scene synthesis as an order-aware sequential learning problem. Forest2Seq organizes the inherently unordered collection of scene objects into structured, ordered hierarchical scene trees and forests. By employing a clustering-based algorithm and a breadth-first traversal, Forest2Seq derives meaningful orderings and utilizes a transformer to generate realistic 3D scenes autoregressively. Experimental results on standard benchmarks demonstrate Forest2Seq's superiority in synthesizing more realistic scenes compared to top-performing baselines, with significant improvements in FID and KL scores. Our additional experiments for downstream tasks and ablation studies also confirm the importance of incorporating order as a prior in 3D scene generation., Comment: ECCV 2024

Fisheye images are categorized fisheye into central and deviated based on the optical center position. Existing rectification methods are limited to central fisheye images, while this paper proposes a novel method that extends to deviated fisheye image rectification. The challenge lies in the variant global distortion distribution pattern caused by the random optical center position. To address this challenge, we propose a distortion vector map (DVM) that measures the degree and direction of local distortion. By learning the DVM, the model can independently identify local distortions at each pixel without relying on global distortion patterns. The model adopts a pre-training and fine-tuning training paradigm. In the pre-training stage, it predicts the distortion vector map and perceives the local distortion features of each pixel. In the fine-tuning stage, it predicts a pixel-wise flow map for deviated fisheye image rectification. We also propose a data augmentation method mixing central, deviated, and distorted-free images. Such data augmentation promotes the model performance in rectifying both central and deviated fisheye images, compared with models trained on single-type fisheye images. Extensive experiments demonstrate the effectiveness and superiority of the proposed method.

Video generation is a challenging yet pivotal task in various industries, such as gaming, e-commerce, and advertising. One significant unresolved aspect within T2V is the effective visualization of text within generated videos. Despite the progress achieved in Text-to-Video~(T2V) generation, current methods still cannot effectively visualize texts in videos directly, as they mainly focus on summarizing semantic scene information, understanding, and depicting actions. While recent advances in image-level visual text generation show promise, transitioning these techniques into the video domain faces problems, notably in preserving textual fidelity and motion coherence. In this paper, we propose an innovative approach termed Text-Animator for visual text video generation. Text-Animator contains a text embedding injection module to precisely depict the structures of visual text in generated videos. Besides, we develop a camera control module and a text refinement module to improve the stability of generated visual text by controlling the camera movement as well as the motion of visualized text. Quantitative and qualitative experimental results demonstrate the superiority of our approach to the accuracy of generated visual text over state-of-the-art video generation methods. The project page can be found at https://laulampaul.github.io/text-animator.html., Comment: Project Page: https://laulampaul.github.io/text-animator.html

Temporal prediction is one of the most important technologies for video compression. Various prediction coding modes are designed in traditional video codecs. Traditional video codecs will adaptively to decide the optimal coding mode according to the prediction quality and reference quality. Recently, learned video codecs have made great progress. However, they ignore the prediction and reference quality adaptation, which leads to incorrect utilization of temporal prediction and reconstruction error propagation. Therefore, in this paper, we first propose a confidence-based prediction quality adaptation (PQA) module to provide explicit discrimination for the spatial and channel-wise prediction quality difference. With this module, the prediction with low quality will be suppressed and that with high quality will be enhanced. The codec can adaptively decide which spatial or channel location of predictions to use. Then, we further propose a reference quality adaptation (RQA) module and an associated repeat-long training strategy to provide dynamic spatially variant filters for diverse reference qualities. With the filters, it is easier for our codec to achieve the target reconstruction quality according to reference qualities, thus reducing the propagation of reconstruction errors. Experimental results show that our codec obtains higher compression performance than the reference software of H.266/VVC and the previous state-of-the-art learned video codecs in both RGB and YUV420 colorspaces.

Recently, there have been efforts to improve the performance in sign language recognition by designing self-supervised learning methods. However, these methods capture limited information from sign pose data in a frame-wise learning manner, leading to sub-optimal solutions. To this end, we propose a simple yet effective self-supervised contrastive learning framework to excavate rich context via spatial-temporal consistency from two distinct perspectives and learn instance discriminative representation for sign language recognition. On one hand, since the semantics of sign language are expressed by the cooperation of fine-grained hands and coarse-grained trunks, we utilize both granularity information and encode them into latent spaces. The consistency between hand and trunk features is constrained to encourage learning consistent representation of instance samples. On the other hand, inspired by the complementary property of motion and joint modalities, we first introduce first-order motion information into sign language modeling. Additionally, we further bridge the interaction between the embedding spaces of both modalities, facilitating bidirectional knowledge transfer to enhance sign language representation. Our method is evaluated with extensive experiments on four public benchmarks, and achieves new state-of-the-art performance with a notable margin. The source code is publicly available at https://github.com/sakura/Code., Comment: Accepted by TIP2023

Spoken language glossification (SLG) aims to translate the spoken language text into the sign language gloss, i.e., a written record of sign language. In this work, we present a framework named $S$emi-$S$upervised $S$poken $L$anguage $G$lossification ($S^3$LG) for SLG. To tackle the bottleneck of limited parallel data in SLG, our $S^3$LG incorporates large-scale monolingual spoken language text into SLG training. The proposed framework follows the self-training structure that iteratively annotates and learns from pseudo labels. Considering the lexical similarity and syntactic difference between sign language and spoken language, our $S^3$LG adopts both the rule-based heuristic and model-based approach for auto-annotation. During training, we randomly mix these complementary synthetic datasets and mark their differences with a special token. As the synthetic data may be less quality, the $S^3$LG further leverages consistency regularization to reduce the negative impact of noise in the synthetic data. Extensive experiments are conducted on public benchmarks to demonstrate the effectiveness of the $S^3$LG. Our code is available at \url{https://github.com/yaohj11/S3LG}., Comment: Accepted to ACL2024 main

Games are widely used as research environments for multi-agent reinforcement learning (MARL), but they pose three significant challenges: limited customization, high computational demands, and oversimplification. To address these issues, we introduce the first publicly available map editor for the popular mobile game Honor of Kings and design a lightweight environment, Mini Honor of Kings (Mini HoK), for researchers to conduct experiments. Mini HoK is highly efficient, allowing experiments to be run on personal PCs or laptops while still presenting sufficient challenges for existing MARL algorithms. We have tested our environment on common MARL algorithms and demonstrated that these algorithms have yet to find optimal solutions within this environment. This facilitates the dissemination and advancement of MARL methods within the research community. Additionally, we hope that more researchers will leverage the Honor of Kings map editor to develop innovative and scientifically valuable new maps. Our code and user manual are available at: https://github.com/tencent-ailab/mini-hok.

Tables contain factual and quantitative data accompanied by various structures and contents that pose challenges for machine comprehension. Previous methods generally design task-specific architectures and objectives for individual tasks, resulting in modal isolation and intricate workflows. In this paper, we present a novel large vision-language model, TabPedia, equipped with a concept synergy mechanism. In this mechanism, all the involved diverse visual table understanding (VTU) tasks and multi-source visual embeddings are abstracted as concepts. This unified framework allows TabPedia to seamlessly integrate VTU tasks, such as table detection, table structure recognition, table querying, and table question answering, by leveraging the capabilities of large language models (LLMs). Moreover, the concept synergy mechanism enables table perception-related and comprehension-related tasks to work in harmony, as they can effectively leverage the needed clues from the corresponding source perception embeddings. Furthermore, to better evaluate the VTU task in real-world scenarios, we establish a new and comprehensive table VQA benchmark, ComTQA, featuring approximately 9,000 QA pairs. Extensive quantitative and qualitative experiments on both table perception and comprehension tasks, conducted across various public benchmarks, validate the effectiveness of our TabPedia. The superior performance further confirms the feasibility of using LLMs for understanding visual tables when all concepts work in synergy. The benchmark ComTQA has been open-sourced at https://huggingface.co/datasets/ByteDance/ComTQA. The source code and model will be released later., Comment: 20 pages, 8 figures

Sign language recognition (SLR) has long been plagued by insufficient model representation capabilities. Although current pre-training approaches have alleviated this dilemma to some extent and yielded promising performance by employing various pretext tasks on sign pose data, these methods still suffer from two primary limitations: 1) Explicit motion information is usually disregarded in previous pretext tasks, leading to partial information loss and limited representation capability. 2) Previous methods focus on the local context of a sign pose sequence, without incorporating the guidance of the global meaning of lexical signs. To this end, we propose a Motion-Aware masked autoencoder with Semantic Alignment (MASA) that integrates rich motion cues and global semantic information in a self-supervised learning paradigm for SLR. Our framework contains two crucial components, i.e., a motion-aware masked autoencoder (MA) and a momentum semantic alignment module (SA). Specifically, in MA, we introduce an autoencoder architecture with a motion-aware masked strategy to reconstruct motion residuals of masked frames, thereby explicitly exploring dynamic motion cues among sign pose sequences. Moreover, in SA, we embed our framework with global semantic awareness by aligning the embeddings of different augmented samples from the input sequence in the shared latent space. In this way, our framework can simultaneously learn local motion cues and global semantic features for comprehensive sign language representation. Furthermore, we conduct extensive experiments to validate the effectiveness of our method, achieving new state-of-the-art performance on four public benchmarks., Comment: Accepted by TCSVT 2024

In recent years, the increasing demand for dynamic 3D assets in design and gaming applications has given rise to powerful generative pipelines capable of synthesizing high-quality 4D objects. Previous methods generally rely on score distillation sampling (SDS) algorithm to infer the unseen views and motion of 4D objects, thus leading to unsatisfactory results with defects like over-saturation and Janus problem. Therefore, inspired by recent progress of video diffusion models, we propose to optimize a 4D representation by explicitly generating multi-view videos from one input image. However, it is far from trivial to handle practical challenges faced by such a pipeline, including dramatic temporal inconsistency, inter-frame geometry and texture diversity, and semantic defects brought by video generation results. To address these issues, we propose DG4D, a novel multi-stage framework that generates high-quality and consistent 4D assets without score distillation. Specifically, collaborative techniques and solutions are developed, including an attention injection strategy to synthesize temporal-consistent multi-view videos, a robust and efficient dynamic reconstruction method based on Gaussian Splatting, and a refinement stage with diffusion prior for semantic restoration. The qualitative results and user preference study demonstrate that our framework outperforms the baselines in generation quality by a considerable margin. Code will be released at \url{https://github.com/jasongzy/EG4D}.

Text-driven human motion generation, as one of the vital tasks in computer-aided content creation, has recently attracted increasing attention. While pioneering research has largely focused on improving numerical performance metrics on given datasets, practical applications reveal a common challenge: existing methods often overfit specific motion expressions in the training data, hindering their ability to generalize to novel descriptions like unseen combinations of motions. This limitation restricts their broader applicability. We argue that the aforementioned problem primarily arises from the scarcity of available motion-text pairs, given the many-to-many nature of text-driven motion generation. To tackle this problem, we formulate text-to-motion generation as a Markov decision process and present \textbf{InstructMotion}, which incorporate the trail and error paradigm in reinforcement learning for generalizable human motion generation. Leveraging contrastive pre-trained text and motion encoders, we delve into optimizing reward design to enable InstructMotion to operate effectively on both paired data, enhancing global semantic level text-motion alignment, and synthetic text-only data, facilitating better generalization to novel prompts without the need for ground-truth motion supervision. Extensive experiments on prevalent benchmarks and also our synthesized unpaired dataset demonstrate that the proposed InstructMotion achieves outstanding performance both quantitatively and qualitatively.

Pre-trained vision-language models (VLMs) have shown remarkable generalization capabilities via prompting, which leverages VLMs as knowledge bases to extract information beneficial for downstream tasks. However, existing methods primarily employ uni-modal prompting, which only engages a uni-modal branch, failing to simultaneously adjust vision-language (V-L) features. Additionally, the one-pass forward pipeline in VLM encoding struggles to align V-L features that have a huge gap. Confronting these challenges, we propose a novel method, Progressive Multi-modal conditional Prompt Tuning (ProMPT). ProMPT exploits a recurrent structure, optimizing and aligning V-L features by iteratively utilizing image and current encoding information. It comprises an initialization and a multi-modal iterative evolution (MIE) module. Initialization is responsible for encoding images and text using a VLM, followed by a feature filter that selects text features similar to image. MIE then facilitates multi-modal prompting through class-conditional vision prompting, instance-conditional text prompting, and feature filtering. In each MIE iteration, vision prompts are obtained from filtered text features via a vision generator, promoting image features to focus more on target object during vision prompting. The encoded image features are fed into a text generator to produce text prompts that are more robust to class shifts. Thus, V-L features are progressively aligned, enabling advance from coarse to exact prediction. Extensive experiments are conducted in three settings to evaluate the efficacy of ProMPT. The results indicate that ProMPT outperforms existing methods on average across all settings, demonstrating its superior generalization and robustness. Code is available at https://github.com/qiuxiaoyu9954/ProMPT.

The advent of Large Multimodal Models (LMMs) has sparked a surge in research aimed at harnessing their remarkable reasoning abilities. However, for understanding text-rich images, challenges persist in fully leveraging the potential of LMMs, and existing methods struggle with effectively processing high-resolution images. In this work, we propose TextCoT, a novel Chain-of-Thought framework for text-rich image understanding. TextCoT utilizes the captioning ability of LMMs to grasp the global context of the image and the grounding capability to examine local textual regions. This allows for the extraction of both global and local visual information, facilitating more accurate question-answering. Technically, TextCoT consists of three stages, including image overview, coarse localization, and fine-grained observation. The image overview stage provides a comprehensive understanding of the global scene information, and the coarse localization stage approximates the image area containing the answer based on the question asked. Then, integrating the obtained global image descriptions, the final stage further examines specific regions to provide accurate answers. Our method is free of extra training, offering immediate plug-and-play functionality. Extensive experiments are conducted on a series of text-rich image question-answering benchmark datasets based on several advanced LMMs, and the results demonstrate the effectiveness and strong generalization ability of our method. Code is available at https://github.com/bzluan/TextCoT.

We present a hybrid-view-based knowledge distillation framework, termed HVDistill, to guide the feature learning of a point cloud neural network with a pre-trained image network in an unsupervised manner. By exploiting the geometric relationship between RGB cameras and LiDAR sensors, the correspondence between the two modalities based on both image-plane view and bird-eye view can be established, which facilitates representation learning. Specifically, the image-plane correspondences can be simply obtained by projecting the point clouds, while the bird-eye-view correspondences can be achieved by lifting pixels to the 3D space with the predicted depths under the supervision of projected point clouds. The image teacher networks provide rich semantics from the image-plane view and meanwhile acquire geometric information from the bird-eye view. Indeed, image features from the two views naturally complement each other and together can ameliorate the learned feature representation of the point cloud student networks. Moreover, with a self-supervised pre-trained 2D network, HVDistill requires neither 2D nor 3D annotations. We pre-train our model on nuScenes dataset and transfer it to several downstream tasks on nuScenes, SemanticKITTI, and KITTI datasets for evaluation. Extensive experimental results show that our method achieves consistent improvements over the baseline trained from scratch and significantly outperforms the existing schemes. Codes are available at git@github.com:zhangsha1024/HVDistill.git.

In embodied vision, Instance ImageGoal Navigation (IIN) requires an agent to locate a specific object depicted in a goal image within an unexplored environment. The primary difficulty of IIN stems from the necessity of recognizing the target object across varying viewpoints and rejecting potential distractors. Existing map-based navigation methods largely adopt the representation form of Bird's Eye View (BEV) maps, which, however, lack the representation of detailed textures in a scene. To address the above issues, we propose a new Gaussian Splatting Navigation (abbreviated as GaussNav) framework for IIN task, which constructs a novel map representation based on 3D Gaussian Splatting (3DGS). The proposed framework enables the agent to not only memorize the geometry and semantic information of the scene, but also retain the textural features of objects. Our GaussNav framework demonstrates a significant leap in performance, evidenced by an increase in Success weighted by Path Length (SPL) from 0.252 to 0.578 on the challenging Habitat-Matterport 3D (HM3D) dataset. Our code will be made publicly available., Comment: conference

Based on multilingual pre-trained models, cross-lingual transfer with prompt learning has shown promising effectiveness, where soft prompt learned in a source language is transferred to target languages for downstream tasks, particularly in the low-resource scenario. To efficiently transfer soft prompt, we propose a novel framework, Multilingual Prompt Translator (MPT), where a multilingual prompt translator is introduced to properly process crucial knowledge embedded in prompt by changing language knowledge while retaining task knowledge. Concretely, we first train prompt in source language and employ translator to translate it into target prompt. Besides, we extend an external corpus as auxiliary data, on which an alignment task for predicted answer probability is designed to convert language knowledge, thereby equipping target prompt with multilingual knowledge. In few-shot settings on XNLI, MPT demonstrates superiority over baselines by remarkable improvements. MPT is more prominent compared with vanilla prompting when transferring to languages quite distinct from source language., Comment: 6 pages, 5 figures, conference

3D Gaussian Splatting (3DGS) has become an emerging tool for dynamic scene reconstruction. However, existing methods focus mainly on extending static 3DGS into a time-variant representation, while overlooking the rich motion information carried by 2D observations, thus suffering from performance degradation and model redundancy. To address the above problem, we propose a novel motion-aware enhancement framework for dynamic scene reconstruction, which mines useful motion cues from optical flow to improve different paradigms of dynamic 3DGS. Specifically, we first establish a correspondence between 3D Gaussian movements and pixel-level flow. Then a novel flow augmentation method is introduced with additional insights into uncertainty and loss collaboration. Moreover, for the prevalent deformation-based paradigm that presents a harder optimization problem, a transient-aware deformation auxiliary module is proposed. We conduct extensive experiments on both multi-view and monocular scenes to verify the merits of our work. Compared with the baselines, our method shows significant superiority in both rendering quality and efficiency.

The task of composed image retrieval (CIR) aims to retrieve images based on the query image and the text describing the users' intent. Existing methods have made great progress with the advanced large vision-language (VL) model in CIR task, however, they generally suffer from two main issues: lack of labeled triplets for model training and difficulty of deployment on resource-restricted environments when deploying the large vision-language model. To tackle the above problems, we propose Image2Sentence based Asymmetric zero-shot composed image retrieval (ISA), which takes advantage of the VL model and only relies on unlabeled images for composition learning. In the framework, we propose a new adaptive token learner that maps an image to a sentence in the word embedding space of VL model. The sentence adaptively captures discriminative visual information and is further integrated with the text modifier. An asymmetric structure is devised for flexible deployment, in which the lightweight model is adopted for the query side while the large VL model is deployed on the gallery side. The global contrastive distillation and the local alignment regularization are adopted for the alignment between the light model and the VL model for CIR task. Our experiments demonstrate that the proposed ISA could better cope with the real retrieval scenarios and further improve retrieval accuracy and efficiency., Comment: ICLR 2024 spotlight

In asymmetric retrieval systems, models with different capacities are deployed on platforms with different computational and storage resources. Despite the great progress, existing approaches still suffer from a dilemma between retrieval efficiency and asymmetric accuracy due to the limited capacity of the lightweight query model. In this work, we propose an Asymmetric Feature Fusion (AFF) paradigm, which advances existing asymmetric retrieval systems by considering the complementarity among different features just at the gallery side. Specifically, it first embeds each gallery image into various features, e.g., local features and global features. Then, a dynamic mixer is introduced to aggregate these features into compact embedding for efficient search. On the query side, only a single lightweight model is deployed for feature extraction. The query model and dynamic mixer are jointly trained by sharing a momentum-updated classifier. Notably, the proposed paradigm boosts the accuracy of asymmetric retrieval without introducing any extra overhead to the query side. Exhaustive experiments on various landmark retrieval datasets demonstrate the superiority of our paradigm.

Asymmetric image retrieval is a task that seeks to balance retrieval accuracy and efficiency by leveraging lightweight and large models for the query and gallery sides, respectively. The key to asymmetric image retrieval is realizing feature compatibility between different models. Despite the great progress, most existing approaches either rely on classifiers inherited from gallery models or simply impose constraints at the instance level, ignoring the structure of embedding space. In this work, we propose a simple yet effective structure similarity preserving method to achieve feature compatibility between query and gallery models. Specifically, we first train a product quantizer offline with the image features embedded by the gallery model. The centroid vectors in the quantizer serve as anchor points in the embedding space of the gallery model to characterize its structure. During the training of the query model, anchor points are shared by the query and gallery models. The relationships between image features and centroid vectors are considered as structure similarities and constrained to be consistent. Moreover, our approach makes no assumption about the existence of any labeled training data and thus can be extended to an unlimited amount of data. Comprehensive experiments on large-scale landmark retrieval demonstrate the effectiveness of our approach. Our code is released at: https://github.com/MCC-WH/SSP.

In this work, we present DeepEraser, an effective deep network for generic text removal. DeepEraser utilizes a recurrent architecture that erases the text in an image via iterative operations. Our idea comes from the process of erasing pencil script, where the text area designated for removal is subject to continuous monitoring and the text is attenuated progressively, ensuring a thorough and clean erasure. Technically, at each iteration, an innovative erasing module is deployed, which not only explicitly aggregates the previous erasing progress but also mines additional semantic context to erase the target text. Through iterative refinements, the text regions are progressively replaced with more appropriate content and finally converge to a relatively accurate status. Furthermore, a custom mask generation strategy is introduced to improve the capability of DeepEraser for adaptive text removal, as opposed to indiscriminately removing all the text in an image. Our DeepEraser is notably compact with only 1.4M parameters and trained in an end-to-end manner. To verify its effectiveness, extensive experiments are conducted on several prevalent benchmarks, including SCUT-Syn, SCUT-EnsText, and Oxford Synthetic text dataset. The quantitative and qualitative results demonstrate the effectiveness of our DeepEraser over the state-of-the-art methods, as well as its strong generalization ability in custom mask text removal. The codes and pre-trained models are available at https://github.com/fh2019ustc/DeepEraser

Knowledge distillation (KD) has been widely adopted to compress large language models (LLMs). Existing KD methods investigate various divergence measures including the Kullback-Leibler (KL), reverse Kullback-Leibler (RKL), and Jensen-Shannon (JS) divergences. However, due to limitations inherent in their assumptions and definitions, these measures fail to deliver effective supervision when few distribution overlap exists between the teacher and the student. In this paper, we show that the aforementioned KL, RKL, and JS divergences respectively suffer from issues of mode-averaging, mode-collapsing, and mode-underestimation, which deteriorates logits-based KD for diverse NLP tasks. We propose the Sinkhorn Knowledge Distillation (SinKD) that exploits the Sinkhorn distance to ensure a nuanced and precise assessment of the disparity between teacher and student distributions. Besides, profit by properties of the Sinkhorn metric, we can get rid of sample-wise KD that restricts the perception of divergence in each teacher-student sample pair. Instead, we propose a batch-wise reformulation to capture geometric intricacies of distributions across samples in the high-dimensional space. Comprehensive evaluation on GLUE and SuperGLUE, in terms of comparability, validity, and generalizability, highlights our superiority over state-of-the-art methods on all kinds of LLMs with encoder-only, encoder-decoder, and decoder-only architectures., Comment: Accepted by COLING 2024

As a new embodied vision task, Instance ImageGoal Navigation (IIN) aims to navigate to a specified object depicted by a goal image in an unexplored environment. The main challenge of this task lies in identifying the target object from different viewpoints while rejecting similar distractors. Existing ImageGoal Navigation methods usually adopt the simple Exploration-Exploitation framework and ignore the identification of specific instance during navigation. In this work, we propose to imitate the human behaviour of ``getting closer to confirm" when distinguishing objects from a distance. Specifically, we design a new modular navigation framework named Instance-aware Exploration-Verification-Exploitation (IEVE) for instance-level image goal navigation. Our method allows for active switching among the exploration, verification, and exploitation actions, thereby facilitating the agent in making reasonable decisions under different situations. On the challenging HabitatMatterport 3D semantic (HM3D-SEM) dataset, our method surpasses previous state-of-the-art work, with a classical segmentation model (0.684 vs. 0.561 success) or a robust model (0.702 vs. 0.561 success)

Video compression performance is closely related to the accuracy of inter prediction. It tends to be difficult to obtain accurate inter prediction for the local video regions with inconsistent motion and occlusion. Traditional video coding standards propose various technologies to handle motion inconsistency and occlusion, such as recursive partitions, geometric partitions, and long-term references. However, existing learned video compression schemes focus on obtaining an overall minimized prediction error averaged over all regions while ignoring the motion inconsistency and occlusion in local regions. In this paper, we propose a spatial decomposition and temporal fusion based inter prediction for learned video compression. To handle motion inconsistency, we propose to decompose the video into structure and detail (SDD) components first. Then we perform SDD-based motion estimation and SDD-based temporal context mining for the structure and detail components to generate short-term temporal contexts. To handle occlusion, we propose to propagate long-term temporal contexts by recurrently accumulating the temporal information of each historical reference feature and fuse them with short-term temporal contexts. With the SDD-based motion model and long short-term temporal contexts fusion, our proposed learned video codec can obtain more accurate inter prediction. Comprehensive experimental results demonstrate that our codec outperforms the reference software of H.266/VVC on all common test datasets for both PSNR and MS-SSIM.

Passive non-line-of-sight (NLOS) imaging has witnessed rapid development in recent years, due to its ability to image objects that are out of sight. The light transport condition plays an important role in this task since changing the conditions will lead to different imaging models. Existing learning-based NLOS methods usually train independent models for different light transport conditions, which is computationally inefficient and impairs the practicality of the models. In this work, we propose NLOS-LTM, a novel passive NLOS imaging method that effectively handles multiple light transport conditions with a single network. We achieve this by inferring a latent light transport representation from the projection image and using this representation to modulate the network that reconstructs the hidden image from the projection image. We train a light transport encoder together with a vector quantizer to obtain the light transport representation. To further regulate this representation, we jointly learn both the reconstruction network and the reprojection network during training. A set of light transport modulation blocks is used to modulate the two jointly trained networks in a multi-scale way. Extensive experiments on a large-scale passive NLOS dataset demonstrate the superiority of the proposed method. The code is available at https://github.com/JerryOctopus/NLOS-LTM.

Recently segment anything model (SAM) has shown powerful segmentation capability and has drawn great attention in computer vision fields. Massive following works have developed various applications based on the pretrained SAM and achieved impressive performance on downstream vision tasks. However, SAM consists of heavy architectures and requires massive computational capacity, which hinders the further application of SAM on computation constrained edge devices. To this end, in this paper we propose a framework to obtain a tiny segment anything model (TinySAM) while maintaining the strong zero-shot performance. We first propose a full-stage knowledge distillation method with hard prompt sampling and hard mask weighting strategy to distill a lightweight student model. We also adapt the post-training quantization to the promptable segmentation task and further reduce the computational cost. Moreover, a hierarchical segmenting everything strategy is proposed to accelerate the everything inference by $2\times$ with almost no performance degradation. With all these proposed methods, our TinySAM leads to orders of magnitude computational reduction and pushes the envelope for efficient segment anything task. Extensive experiments on various zero-shot transfer tasks demonstrate the significantly advantageous performance of our TinySAM against counterpart methods. Pre-trained models and codes are available at https://github.com/xinghaochen/TinySAM and https://gitee.com/mindspore/models/tree/master/research/cv/TinySAM.

The utilization of artificial intelligence (AI) in card games has been a well-explored subject within AI research for an extensive period. Recent advancements have propelled AI programs to showcase expertise in intricate card games such as Mahjong, DouDizhu, and Texas Hold'em. In this work, we aim to develop an AI program for an exceptionally complex and popular card game called GuanDan. This game involves four players engaging in both competitive and cooperative play throughout a long process to upgrade their level, posing great challenges for AI due to its expansive state and action space, long episode length, and complex rules. Employing reinforcement learning techniques, specifically Deep Monte Carlo (DMC), and a distributed training framework, we first put forward an AI program named DanZero for this game. Evaluation against baseline AI programs based on heuristic rules highlights the outstanding performance of our bot. Besides, in order to further enhance the AI's capabilities, we apply policy-based reinforcement learning algorithm to GuanDan. To address the challenges arising from the huge action space, which will significantly impact the performance of policy-based algorithms, we adopt the pre-trained model to facilitate the training process and the achieved AI program manages to achieve a superior performance., Comment: arXiv admin note: text overlap with arXiv:2210.17087

In the field of document understanding, significant advances have been made in the fine-tuning of Multimodal Large Language Models (MLLMs) with instruction-following data. Nevertheless, the potential of text-grounding capability within text-rich scenarios remains underexplored. In this paper, we present a text-grounding document understanding model, termed TGDoc, which addresses this deficiency by enhancing MLLMs with the ability to discern the spatial positioning of text within images. Empirical evidence suggests that text-grounding improves the model's interpretation of textual content, thereby elevating its proficiency in comprehending text-rich images. Specifically, we compile a dataset containing 99K PowerPoint presentations sourced from the internet. We formulate instruction tuning tasks including text detection, recognition, and spotting to facilitate the cohesive alignment between the visual encoder and large language model. Moreover, we curate a collection of text-rich images and prompt the text-only GPT-4 to generate 12K high-quality conversations, featuring textual locations within text-rich scenarios. By integrating text location data into the instructions, TGDoc is adept at discerning text locations during the visual question process. Extensive experiments demonstrate that our method achieves state-of-the-art performance across multiple text-rich benchmarks, validating the effectiveness of our method.

This work presents DocPedia, a novel large multimodal model (LMM) for versatile OCR-free document understanding, capable of parsing images up to 2,560$\times$2,560 resolution. Unlike existing work either struggle with high-resolution documents or give up the large language model thus vision or language ability constrained, our DocPedia directly processes visual input in the frequency domain rather than the pixel space. The unique characteristic enables DocPedia to capture a greater amount of visual and textual information using a limited number of visual tokens. To consistently enhance both perception and comprehension abilities of our model, we develop a dual-stage training strategy and enrich instructions/annotations of all training tasks covering multiple document types. Extensive quantitative and qualitative experiments conducted on various publicly available benchmarks confirm the mutual benefits of jointly learning perception and comprehension tasks. The results provide further evidence of the effectiveness and superior performance of our DocPedia over other methods.

Pre-training is playing an increasingly important role in learning generic feature representation for Person Re-identification (ReID). We argue that a high-quality ReID representation should have three properties, namely, multi-level awareness, occlusion robustness, and cross-region invariance. To this end, we propose a simple yet effective pre-training framework, namely PersonMAE, which involves two core designs into masked autoencoders to better serve the task of Person Re-ID. 1) PersonMAE generates two regions from the given image with RegionA as the input and \textit{RegionB} as the prediction target. RegionA is corrupted with block-wise masking to mimic common occlusion in ReID and its remaining visible parts are fed into the encoder. 2) Then PersonMAE aims to predict the whole RegionB at both pixel level and semantic feature level. It encourages its pre-trained feature representations with the three properties mentioned above. These properties make PersonMAE compatible with downstream Person ReID tasks, leading to state-of-the-art performance on four downstream ReID tasks, i.e., supervised (holistic and occluded setting), and unsupervised (UDA and USL setting). Notably, on the commonly adopted supervised setting, PersonMAE with ViT-B backbone achieves 79.8% and 69.5% mAP on the MSMT17 and OccDuke datasets, surpassing the previous state-of-the-art by a large margin of +8.0 mAP, and +5.3 mAP, respectively.

Single-image shadow removal is a significant task that is still unresolved. Most existing deep learning-based approaches attempt to remove the shadow directly, which can not deal with the shadow well. To handle this issue, we consider removing the shadow in a coarse-to-fine fashion and propose a simple but effective Progressive Recurrent Network (PRNet). The network aims to remove the shadow progressively, enabing us to flexibly adjust the number of iterations to strike a balance between performance and time. Our network comprises two parts: shadow feature extraction and progressive shadow removal. Specifically, the first part is a shallow ResNet which constructs the representations of the input shadow image on its original size, preventing the loss of high-frequency details caused by the downsampling operation. The second part has two critical components: the re-integration module and the update module. The proposed re-integration module can fully use the outputs of the previous iteration, providing input for the update module for further shadow removal. In this way, the proposed PRNet makes the whole process more concise and only uses 29% network parameters than the best published method. Extensive experiments on the three benchmarks, ISTD, ISTD+, and SRD, demonstrate that our method can effectively remove shadows and achieve superior performance.

While deep reinforcement learning (RL) has been demonstrated effective in solving complex control tasks, sample efficiency remains a key challenge due to the large amounts of data required for remarkable performance. Existing research explores the application of representation learning for data-efficient RL, e.g., learning predictive representations by predicting long-term future states. However, many existing methods do not fully exploit the structural information inherent in sequential state signals, which can potentially improve the quality of long-term decision-making but is difficult to discern in the time domain. To tackle this problem, we propose State Sequences Prediction via Fourier Transform (SPF), a novel method that exploits the frequency domain of state sequences to extract the underlying patterns in time series data for learning expressive representations efficiently. Specifically, we theoretically analyze the existence of structural information in state sequences, which is closely related to policy performance and signal regularity, and then propose to predict the Fourier transform of infinite-step future state sequences to extract such information. One of the appealing features of SPF is that it is simple to implement while not requiring storage of infinite-step future states as prediction targets. Experiments demonstrate that the proposed method outperforms several state-of-the-art algorithms in terms of both sample efficiency and performance.

Recent progresses on self-supervised 3D human action representation learning are largely attributed to contrastive learning. However, in conventional contrastive frameworks, the rich complementarity between different skeleton modalities remains under-explored. Moreover, optimized with distinguishing self-augmented samples, models struggle with numerous similar positive instances in the case of limited action categories. In this work, we tackle the aforementioned problems by introducing a general Inter- and Intra-modal Mutual Distillation (I$^2$MD) framework. In I$^2$MD, we first re-formulate the cross-modal interaction as a Cross-modal Mutual Distillation (CMD) process. Different from existing distillation solutions that transfer the knowledge of a pre-trained and fixed teacher to the student, in CMD, the knowledge is continuously updated and bidirectionally distilled between modalities during pre-training. To alleviate the interference of similar samples and exploit their underlying contexts, we further design the Intra-modal Mutual Distillation (IMD) strategy, In IMD, the Dynamic Neighbors Aggregation (DNA) mechanism is first introduced, where an additional cluster-level discrimination branch is instantiated in each modality. It adaptively aggregates highly-correlated neighboring features, forming local cluster-level contrasting. Mutual distillation is then performed between the two branches for cross-level knowledge exchange. Extensive experiments on three datasets show that our approach sets a series of new records., Comment: submitted to IJCV. arXiv admin note: substantial text overlap with arXiv:2208.12448

Large-scale LP problems from industry usually contain much redundancy that severely hurts the efficiency and reliability of solving LPs, making presolve (i.e., the problem simplification module) one of the most critical components in modern LP solvers. However, how to design high-quality presolve routines -- that is, the program determining (P1) which presolvers to select, (P2) in what order to execute, and (P3) when to stop -- remains a highly challenging task due to the extensive requirements on expert knowledge and the large search space. Due to the sequential decision property of the task and the lack of expert demonstrations, we propose a simple and efficient reinforcement learning (RL) framework -- namely, reinforcement learning for presolve (RL4Presolve) -- to tackle (P1)-(P3) simultaneously. Specifically, we formulate the routine design task as a Markov decision process and propose an RL framework with adaptive action sequences to generate high-quality presolve routines efficiently. Note that adaptive action sequences help learn complex behaviors efficiently and adapt to various benchmarks. Experiments on two solvers (open-source and commercial) and eight benchmarks (real-world and synthetic) demonstrate that RL4Presolve significantly and consistently improves the efficiency of solving large-scale LPs, especially on benchmarks from industry. Furthermore, we optimize the hard-coded presolve routines in LP solvers by extracting rules from learned policies for simple and efficient deployment to Huawei's supply chain. The results show encouraging economic and academic potential for incorporating machine learning to modern solvers.

As vehicular communication and networking technologies continue to advance, infrastructure-based roadside perception emerges as a pivotal tool for connected automated vehicle (CAV) applications. Due to their elevated positioning, roadside sensors, including cameras and lidars, often enjoy unobstructed views with diminished object occlusion. This provides them a distinct advantage over onboard perception, enabling more robust and accurate detection of road objects. This paper presents MSight, a cutting-edge roadside perception system specifically designed for CAVs. MSight offers real-time vehicle detection, localization, tracking, and short-term trajectory prediction. Evaluations underscore the system's capability to uphold lane-level accuracy with minimal latency, revealing a range of potential applications to enhance CAV safety and efficiency. Presently, MSight operates 24/7 at a two-lane roundabout in the City of Ann Arbor, Michigan., Comment: Submitted to IEEE T-ITS

In recent years, large-scale language models (LLMs) have gained attention for their impressive text generation capabilities. However, these models often face the challenge of "hallucination," which undermines their reliability. In this study, we introduce an uncertainty-aware in-context learning framework to empower the model to enhance or reject its output in response to uncertainty. Human-defined methods for estimating uncertainty typically assume that "uncertainty is lower when the model's response is correct compared to when it is incorrect." However, setting a precise threshold to distinguish correctness is challenging. Therefore, we introduce uncertainty information as an intermediary variable that implicitly influences the model's behavior. Our innovative uncertainty-aware in-context learning framework involves fine-tuning the LLM using a calibration dataset. Our aim is to improve the model's responses by filtering out answers with high uncertainty while considering the model's knowledge limitations. We evaluate the model's knowledge by examining multiple responses to the same question for the presence of a correct answer. When the model lacks relevant knowledge, the response should indicate that the question cannot be answered. Conversely, when the model has relevant knowledge, the response should provide the correct answer. Extensive experiments confirm the effectiveness of our framework, leading to two key findings. First, the logit output values of the LLM partly reflect inherent uncertainty. Second, our model autonomously recognizes uncertainty, resulting in improved responses.

This paper presents IP-SLT, a simple yet effective framework for sign language translation (SLT). Our IP-SLT adopts a recurrent structure and enhances the semantic representation (prototype) of the input sign language video via an iterative refinement manner. Our idea mimics the behavior of human reading, where a sentence can be digested repeatedly, till reaching accurate understanding. Technically, IP-SLT consists of feature extraction, prototype initialization, and iterative prototype refinement. The initialization module generates the initial prototype based on the visual feature extracted by the feature extraction module. Then, the iterative refinement module leverages the cross-attention mechanism to polish the previous prototype by aggregating it with the original video feature. Through repeated refinement, the prototype finally converges to a more stable and accurate state, leading to a fluent and appropriate translation. In addition, to leverage the sequential dependence of prototypes, we further propose an iterative distillation loss to compress the knowledge of the final iteration into previous ones. As the autoregressive decoding process is executed only once in inference, our IP-SLT is ready to improve various SLT systems with acceptable overhead. Extensive experiments are conducted on public benchmarks to demonstrate the effectiveness of the IP-SLT., Comment: Accepted by ICCV 2023

In the era of Large Language Models (LLMs), tremendous strides have been made in the field of multimodal understanding. However, existing advanced algorithms are limited to effectively utilizing the immense representation capabilities and rich world knowledge inherent to these large pre-trained models, and the beneficial connections among tasks within the context of text-rich scenarios have not been sufficiently explored. In this work, we introduce UniDoc, a novel multimodal model equipped with text detection and recognition capabilities, which are deficient in existing approaches. Moreover, UniDoc capitalizes on the beneficial interactions among tasks to enhance the performance of each individual task. To implement UniDoc, we perform unified multimodal instruct tuning on the contributed large-scale instruction following datasets. Quantitative and qualitative experimental results show that UniDoc sets state-of-the-art scores across multiple challenging benchmarks. To the best of our knowledge, this is the first large multimodal model capable of simultaneous text detection, recognition, spotting, and understanding.

In fisheye images, rich distinct distortion patterns are regularly distributed in the image plane. These distortion patterns are independent of the visual content and provide informative cues for rectification. To make the best of such rectification cues, we introduce SimFIR, a simple framework for fisheye image rectification based on self-supervised representation learning. Technically, we first split a fisheye image into multiple patches and extract their representations with a Vision Transformer (ViT). To learn fine-grained distortion representations, we then associate different image patches with their specific distortion patterns based on the fisheye model, and further subtly design an innovative unified distortion-aware pretext task for their learning. The transfer performance on the downstream rectification task is remarkably boosted, which verifies the effectiveness of the learned representations. Extensive experiments are conducted, and the quantitative and qualitative results demonstrate the superiority of our method over the state-of-the-art algorithms as well as its strong generalization ability on real-world fisheye images., Comment: Accepted to ICCV 2023

Visual storytelling aims to generate a narrative based on a sequence of images, necessitating both vision-language alignment and coherent story generation. Most existing solutions predominantly depend on paired image-text training data, which can be costly to collect and challenging to scale. To address this, we formulate visual storytelling as a visual-conditioned story generation problem and propose a text-only training method that separates the learning of cross-modality alignment and story generation. Our approach specifically leverages the cross-modality pre-trained CLIP model to integrate visual control into a story generator, trained exclusively on text data. Moreover, we devise a training-free visual condition planner that accounts for the temporal structure of the input image sequence while balancing global and local visual content. The distinctive advantage of requiring only text data for training enables our method to learn from external text story data, enhancing the generalization capability of visual storytelling. We conduct extensive experiments on the VIST benchmark, showcasing the effectiveness of our approach in both in-domain and cross-domain settings. Further evaluations on expression diversity and human assessment underscore the superiority of our method in terms of informativeness and robustness., Comment: ACM MM 2023

Controllable video generation has gained significant attention in recent years. However, two main limitations persist: Firstly, most existing works focus on either text, image, or trajectory-based control, leading to an inability to achieve fine-grained control in videos. Secondly, trajectory control research is still in its early stages, with most experiments being conducted on simple datasets like Human3.6M. This constraint limits the models' capability to process open-domain images and effectively handle complex curved trajectories. In this paper, we propose DragNUWA, an open-domain diffusion-based video generation model. To tackle the issue of insufficient control granularity in existing works, we simultaneously introduce text, image, and trajectory information to provide fine-grained control over video content from semantic, spatial, and temporal perspectives. To resolve the problem of limited open-domain trajectory control in current research, We propose trajectory modeling with three aspects: a Trajectory Sampler (TS) to enable open-domain control of arbitrary trajectories, a Multiscale Fusion (MF) to control trajectories in different granularities, and an Adaptive Training (AT) strategy to generate consistent videos following trajectories. Our experiments validate the effectiveness of DragNUWA, demonstrating its superior performance in fine-grained control in video generation. The homepage link is \url{https://www.microsoft.com/en-us/research/project/dragnuwa/}

In 3D human action recognition, limited supervised data makes it challenging to fully tap into the modeling potential of powerful networks such as transformers. As a result, researchers have been actively investigating effective self-supervised pre-training strategies. In this work, we show that instead of following the prevalent pretext task to perform masked self-component reconstruction in human joints, explicit contextual motion modeling is key to the success of learning effective feature representation for 3D action recognition. Formally, we propose the Masked Motion Prediction (MAMP) framework. To be specific, the proposed MAMP takes as input the masked spatio-temporal skeleton sequence and predicts the corresponding temporal motion of the masked human joints. Considering the high temporal redundancy of the skeleton sequence, in our MAMP, the motion information also acts as an empirical semantic richness prior that guide the masking process, promoting better attention to semantically rich temporal regions. Extensive experiments on NTU-60, NTU-120, and PKU-MMD datasets show that the proposed MAMP pre-training substantially improves the performance of the adopted vanilla transformer, achieving state-of-the-art results without bells and whistles. The source code of our MAMP is available at https://github.com/maoyunyao/MAMP., Comment: To appear in ICCV 2023