Wenjing Wang

Junchen Zhu, Huan Yang, Huiguo He et al. · microsoft-research

In this paper, we present MovieFactory, a powerful framework to generate cinematic-picture (3072$\times$1280), film-style (multi-scene), and multi-modality (sounding) movies on the demand of natural languages. As the first fully automated movie generation model to the best of our knowledge, our approach empowers users to create captivating movies with smooth transitions using simple text inputs, surpassing existing methods that produce soundless videos limited to a single scene of modest quality. To facilitate this distinctive functionality, we leverage ChatGPT to expand user-provided text into detailed sequential scripts for movie generation. Then we bring scripts to life visually and acoustically through vision generation and audio retrieval. To generate videos, we extend the capabilities of a pretrained text-to-image diffusion model through a two-stage process. Firstly, we employ spatial finetuning to bridge the gap between the pretrained image model and the new video dataset. Subsequently, we introduce temporal learning to capture object motion. In terms of audio, we leverage sophisticated retrieval models to select and align audio elements that correspond to the plot and visual content of the movie. Extensive experiments demonstrate that our MovieFactory produces movies with realistic visuals, diverse scenes, and seamlessly fitting audio, offering users a novel and immersive experience. Generated samples can be found in YouTube or Bilibili (1080P).

Yiyang Ma, Huan Yang, Wenjing Wang et al. · microsoft-research

Language-guided image generation has achieved great success nowadays by using diffusion models. However, texts can be less detailed to describe highly-specific subjects such as a particular dog or a certain car, which makes pure text-to-image generation not accurate enough to satisfy user requirements. In this work, we present a novel Unified Multi-Modal Latent Diffusion (UMM-Diffusion) which takes joint texts and images containing specified subjects as input sequences and generates customized images with the subjects. To be more specific, both input texts and images are encoded into one unified multi-modal latent space, in which the input images are learned to be projected to pseudo word embedding and can be further combined with text to guide image generation. Besides, to eliminate the irrelevant parts of the input images such as background or illumination, we propose a novel sampling technique of diffusion models used by the image generator which fuses the results guided by multi-modal input and pure text input. By leveraging the large-scale pre-trained text-to-image generator and the designed image encoder, our method is able to generate high-quality images with complex semantics from both aspects of input texts and images.

Junchen Zhu, Huan Yang, Wenjing Wang et al. · microsoft-research

Videos for mobile devices become the most popular access to share and acquire information recently. For the convenience of users' creation, in this paper, we present a system, namely MobileVidFactory, to automatically generate vertical mobile videos where users only need to give simple texts mainly. Our system consists of two parts: basic and customized generation. In the basic generation, we take advantage of the pretrained image diffusion model, and adapt it to a high-quality open-domain vertical video generator for mobile devices. As for the audio, by retrieving from our big database, our system matches a suitable background sound for the video. Additionally to produce customized content, our system allows users to add specified screen texts to the video for enriching visual expression, and specify texts for automatic reading with optional voices as they like.

Rundong Luo, Wenjing Wang, Wenhan Yang et al.

Low-light conditions not only hamper human visual experience but also degrade the model's performance on downstream vision tasks. While existing works make remarkable progress on day-night domain adaptation, they rely heavily on domain knowledge derived from the task-specific nighttime dataset. This paper challenges a more complicated scenario with border applicability, i.e., zero-shot day-night domain adaptation, which eliminates reliance on any nighttime data. Unlike prior zero-shot adaptation approaches emphasizing either image-level translation or model-level adaptation, we propose a similarity min-max paradigm that considers them under a unified framework. On the image level, we darken images towards minimum feature similarity to enlarge the domain gap. Then on the model level, we maximize the feature similarity between the darkened images and their normal-light counterparts for better model adaptation. To the best of our knowledge, this work represents the pioneering effort in jointly optimizing both aspects, resulting in a significant improvement of model generalizability. Extensive experiments demonstrate our method's effectiveness and broad applicability on various nighttime vision tasks, including classification, semantic segmentation, visual place recognition, and video action recognition. Code and pre-trained models are available at https://red-fairy.github.io/ZeroShotDayNightDA-Webpage/.

Wenjing Wang, Lilang Lin, Zejia Fan et al.

With the progress of Mars exploration, numerous Mars image data are collected and need to be analyzed. However, due to the imbalance and distortion of Martian data, the performance of existing computer vision models is unsatisfactory. In this paper, we introduce a semi-supervised framework for machine vision on Mars and try to resolve two specific tasks: classification and segmentation. Contrastive learning is a powerful representation learning technique. However, there is too much information overlap between Martian data samples, leading to a contradiction between contrastive learning and Martian data. Our key idea is to reconcile this contradiction with the help of annotations and further take advantage of unlabeled data to improve performance. For classification, we propose to ignore inner-class pairs on labeled data as well as neglect negative pairs on unlabeled data, forming supervised inter-class contrastive learning and unsupervised similarity learning. For segmentation, we extend supervised inter-class contrastive learning into an element-wise mode and use online pseudo labels for supervision on unlabeled areas. Experimental results show that our learning strategies can improve the classification and segmentation models by a large margin and outperform state-of-the-art approaches.

Wenjing Wang, Zhengbo Xu, Haofeng Huang et al.

Low light conditions not only degrade human visual experience, but also reduce the performance of downstream machine analytics. Although many works have been designed for low-light enhancement or domain adaptive machine analytics, the former considers less on high-level vision, while the latter neglects the potential of image-level signal adjustment. How to restore underexposed images/videos from the perspective of machine vision has long been overlooked. In this paper, we are the first to propose a learnable illumination enhancement model for high-level vision. Inspired by real camera response functions, we assume that the illumination enhancement function should be a concave curve, and propose to satisfy this concavity through discrete integral. With the intention of adapting illumination from the perspective of machine vision without task-specific annotated data, we design an asymmetric cross-domain self-supervised training strategy. Our model architecture and training designs mutually benefit each other, forming a powerful unsupervised normal-to-low light adaptation framework. Comprehensive experiments demonstrate that our method surpasses existing low-light enhancement and adaptation methods and shows superior generalization on various low-light vision tasks, including classification, detection, action recognition, and optical flow estimation. Project website: https://daooshee.github.io/SACC-Website/

Jiahang Zhang, Lilang Lin, Zejia Fan et al.

Deep learning has become a powerful tool for Mars exploration. Mars terrain semantic segmentation is an important Martian vision task, which is the base of rover autonomous planning and safe driving. However, there is a lack of sufficient detailed and high-confidence data annotations, which are exactly required by most deep learning methods to obtain a good model. To address this problem, we propose our solution from the perspective of joint data and method design. We first present a newdataset S5Mars for Semi-SuperviSed learning on Mars Semantic Segmentation, which contains 6K high-resolution images and is sparsely annotated based on confidence, ensuring the high quality of labels. Then to learn from this sparse data, we propose a semi-supervised learning (SSL) framework for Mars image semantic segmentation, to learn representations from limited labeled data. Different from the existing SSL methods which are mostly targeted at the Earth image data, our method takes into account Mars data characteristics. Specifically, we first investigate the impact of current widely used natural image augmentations on Mars images. Based on the analysis, we then proposed two novel and effective augmentations for SSL of Mars segmentation, AugIN and SAM-Mix, which serve as strong augmentations to boost the model performance. Meanwhile, to fully leverage the unlabeled data, we introduce a soft-to-hard consistency learning strategy, learning from different targets based on prediction confidence. Experimental results show that our method can outperform state-of-the-art SSL approaches remarkably. Our proposed dataset is available at https://jhang2020.github.io/S5Mars.github.io/.

Dingming Liu, Wenjing Wang, Chen Li et al.

Video object removal aims to eliminate target objects from videos while plausibly completing missing regions and preserving spatio-temporal consistency. Although diffusion models have recently advanced this task, it remains challenging to remove object-induced side effects (e.g., shadows, reflections, and illumination changes) without compromising overall coherence. This limitation stems from the insufficient physical and semantic understanding of the target object and its interactions with the scene. In this paper, we propose to introduce understanding into erasing from two complementary perspectives. Externally, we introduce a distillation scheme that transfers the relationships between objects and their induced effects from vision foundation models to video diffusion models. Internally, we propose a framewise context cross-attention mechanism that grounds each denoising block in informative, unmasked context surrounding the target region. External and internal guidance jointly enable our model to understand the target object, its induced effects, and the global background context, resulting in clear and coherent object removal. Extensive experiments demonstrate our state-of-the-art performance, and we establish the first real-world benchmark for video object removal to facilitate future research and community progress. Our code, data, and models are available at: https://github.com/WeChatCV/UnderEraser.

Yiwen Zhu, Joyce Cahoon, Anna Pavlenko et al.

Complex operational workflows coordinating personnel, tools, and information are central to enterprise operations, yet end-to-end automation remains challenging due to extensive requirements for human inputs and the inability to adapt over time. We present GraphMind, an end-to-end system that constructs, executes, and evolves action-centric workflow graphs without human effort. The system operates in three phases. First, a scalable offline pipeline extracts structured workflow graphs from large volumes of human resolution traces, capturing problems, actions, and their causal relationships. Second, an online multi-agent traversal engine navigates the graph to dynamically construct and execute workflows, combining graph-guided retrieval with LLM-driven reasoning at each step. Third, Adaptive Traversal Reinforcement (ATR) reinforces successful traversal paths and decays stale elements. This closed-loop mechanism enables the graph to self-optimize and adapt to shifting operational conditions. GraphMind has been deployed across four production cloud database services for incident investigation. Evaluated on production data, the system substantially outperforms a Trace-RAG baseline in mitigation reach, groundedness, and diagnostic throughput, scoring 4.95/5 in blind expert review. The ATR layer provides further gains across all metrics, demonstrating that workflow graphs can learn and improve from execution-derived feedback.

Yuhang Zhang, Yue Yao, Xuannan Liu et al.

Facial expression recognition (FER) models are typically trained on datasets with a fixed number of seven basic classes. However, recent research works point out that there are far more expressions than the basic ones. Thus, when these models are deployed in the real world, they may encounter unknown classes, such as compound expressions that cannot be classified into existing basic classes. To address this issue, we propose the open-set FER task for the first time. Though there are many existing open-set recognition methods, we argue that they do not work well for open-set FER because FER data are all human faces with very small inter-class distances, which makes the open-set samples very similar to close-set samples. In this paper, we are the first to transform the disadvantage of small inter-class distance into an advantage by proposing a new way for open-set FER. Specifically, we find that small inter-class distance allows for sparsely distributed pseudo labels of open-set samples, which can be viewed as symmetric noisy labels. Based on this novel observation, we convert the open-set FER to a noisy label detection problem. We further propose a novel method that incorporates attention map consistency and cycle training to detect the open-set samples. Extensive experiments on various FER datasets demonstrate that our method clearly outperforms state-of-the-art open-set recognition methods by large margins. Code is available at https://github.com/zyh-uaiaaaa.

Wenjing Wang, Wenxuan Wang, Songning Lai

While deep learning has significantly advanced accident anticipation, the robustness of these safety-critical systems against real-world perturbations remains a major challenge. We reveal that state-of-the-art models like CRASH, despite their high performance, exhibit significant instability in predictions and latent representations when faced with minor input perturbations, posing serious reliability risks. To address this, we introduce SECURE - Stable Early Collision Understanding Robust Embeddings, a framework that formally defines and enforces model robustness. SECURE is founded on four key attributes: consistency and stability in both prediction space and latent feature space. We propose a principled training methodology that fine-tunes a baseline model using a multi-objective loss, which minimizes divergence from a reference model and penalizes sensitivity to adversarial perturbations. Experiments on DAD and CCD datasets demonstrate that our approach not only significantly enhances robustness against various perturbations but also improves performance on clean data, achieving new state-of-the-art results.

Wenjing Wang, Huan Yang, Jianlong Fu et al.

Understanding illumination and reducing the need for supervision pose a significant challenge in low-light enhancement. Current approaches are highly sensitive to data usage during training and illumination-specific hyper-parameters, limiting their ability to handle unseen scenarios. In this paper, we propose a new zero-reference low-light enhancement framework trainable solely with normal light images. To accomplish this, we devise an illumination-invariant prior inspired by the theory of physical light transfer. This prior serves as the bridge between normal and low-light images. Then, we develop a prior-to-image framework trained without low-light data. During testing, this framework is able to restore our illumination-invariant prior back to images, automatically achieving low-light enhancement. Within this framework, we leverage a pretrained generative diffusion model for model ability, introduce a bypass decoder to handle detail distortion, as well as offer a lightweight version for practicality. Extensive experiments demonstrate our framework's superiority in various scenarios as well as good interpretability, robustness, and efficiency. Code is available on our project homepage: http://daooshee.github.io/QuadPrior-Website/

Wenjing Wang, Huan Yang, Zixi Tuo et al.

With the explosive popularity of AI-generated content (AIGC), video generation has recently received a lot of attention. Generating videos guided by text instructions poses significant challenges, such as modeling the complex relationship between space and time, and the lack of large-scale text-video paired data. Existing text-video datasets suffer from limitations in both content quality and scale, or they are not open-source, rendering them inaccessible for study and use. For model design, previous approaches extend pretrained text-to-image generation models by adding temporal 1D convolution/attention modules for video generation. However, these approaches overlook the importance of jointly modeling space and time, inevitably leading to temporal distortions and misalignment between texts and videos. In this paper, we propose a novel approach that strengthens the interaction between spatial and temporal perceptions. In particular, we utilize a swapped cross-attention mechanism in 3D windows that alternates the "query" role between spatial and temporal blocks, enabling mutual reinforcement for each other. Moreover, to fully unlock model capabilities for high-quality video generation and promote the development of the field, we curate a large-scale and open-source video dataset called HD-VG-130M. This dataset comprises 130 million text-video pairs from the open-domain, ensuring high-definition, widescreen and watermark-free characters. A smaller-scale yet more meticulously cleaned subset further enhances the data quality, aiding models in achieving superior performance. Experimental quantitative and qualitative results demonstrate the superiority of our approach in terms of per-frame quality, temporal correlation, and text-video alignment, with clear margins.

Ruiying Sun, Wenjing Wang, Qinhan Chen et al.

Real-world mathematical modeling is inherently an experiential and collaborative endeavor. Domain experts rarely solve complex problems from scratch; instead, they draw upon analogies from historical cases and subject their hypotheses to rigorous peer scrutiny. However, autonomous agents powered by Large Language Models predominantly rely on isolated reasoning paradigms, frequently generating plausible but fundamentally flawed models due to a lack of domain grounding and adversarial verification. To address these limitations, we propose Sci-Mind, a novel framework that mirrors the human scientific discovery process. Sci-Mind integrates Experiential Memory Recall to retrieve executable code snippets and modeling paradigm descriptors, grounding abstract reasoning in historical solutions. Subsequently, it employs an Adversarial Cognitive Dialectic where a Theorist optimizing mathematical coherence and a Pragmatist enforcing data feasibility debate through competing objectives to prune elegant but infeasible formulations. A Self-Validating Execution Strategy further ensures blueprint consistency through formal predicates before code generation, achieving fully autonomous execution. Extensive experiments on the MM-Bench and EngiBench benchmarks demonstrate that Sci-Mind significantly outperforms leading autonomous agents in both modeling rigorousness and code executability.

Yingjie Chen, Shilun Lin, Cai Xing et al.

Recent advances have demonstrated compelling capabilities in synthesizing real individuals into generated videos, reflecting the growing demand for identity-aware content creation. Nevertheless, an openly accessible framework enabling fine-grained control over facial appearance and voice timbre across multiple identities remains unavailable. In this work, we present a unified and scalable framework for identity-aware joint audio-video generation, enabling high-fidelity and consistent personalization. Specifically, we introduce a data curation pipeline that automatically extracts identity-bearing information with paired annotations across audio and visual modalities, covering diverse scenarios from single-subject to multi-subject interactions. We further propose a flexible and scalable identity injection mechanism for single- and multi-subject scenarios, in which both facial appearance and vocal timbre act as identity-bearing control signals. Moreover, in light of modality disparity, we design a multi-stage training strategy to accelerate convergence and enforce cross-modal coherence. Experiments demonstrate the superiority of the proposed framework. For more details and qualitative results, please refer to our webpage: \href{https://chen-yingjie.github.io/projects/Identity-as-Presence}{Identity-as-Presence}.

Bowen Xue, Qixin Yan, Wenjing Wang et al.

Generating high-fidelity human videos that match user-specified identities is important yet challenging in the field of generative AI. Existing methods often rely on an excessive number of training parameters and lack compatibility with other AIGC tools. In this paper, we propose Stand-In, a lightweight and plug-and-play framework for identity preservation in video generation. Specifically, we introduce a conditional image branch into the pre-trained video generation model. Identity control is achieved through restricted self-attentions with conditional position mapping, and can be learned quickly with only 2000 pairs. Despite incorporating and training just $\sim$1% additional parameters, our framework achieves excellent results in video quality and identity preservation, outperforming other full-parameter training methods. Moreover, our framework can be seamlessly integrated for other tasks, such as subject-driven video generation, pose-referenced video generation, stylization, and face swapping.

Yiwen Zhu, Mathieu Demarne, Kai Deng et al.

Software engineers frequently grapple with the challenge of accessing disparate documentation and telemetry data, including TroubleShooting Guides (TSGs), incident reports, code repositories, and various internal tools developed by multiple stakeholders. While on-call duties are inevitable, incident resolution becomes even more daunting due to the obscurity of legacy sources and the pressures of strict time constraints. To enhance the efficiency of on-call engineers (OCEs) and streamline their daily workflows, we introduced DECO-a comprehensive framework for developing, deploying, and managing enterprise-grade copilots tailored to improve productivity in engineering routines. This paper details the design and implementation of the DECO framework, emphasizing its innovative NL2SearchQuery functionality and a lightweight agentic framework. These features support efficient and customized retrieval-augmented-generation (RAG) algorithms that not only extract relevant information from diverse sources but also select the most pertinent skills in response to user queries. This enables the addressing of complex technical questions and provides seamless, automated access to internal resources. Additionally, DECO incorporates a robust mechanism for converting unstructured incident logs into user-friendly, structured guides, effectively bridging the documentation gap. Since its launch in September 2023, DECO has demonstrated its effectiveness through widespread adoption, enabling tens of thousands of interactions and engaging hundreds of monthly active users (MAU) across dozens of organizations within the company.

Haotian Dong, Wenjing Wang, Chen Li et al.

RGBA video generation, which includes an alpha channel to represent transparency, is gaining increasing attention across a wide range of applications. However, existing methods often neglect visual quality, limiting their practical usability. In this paper, we propose Wan-Alpha, a new framework that generates transparent videos by learning both RGB and alpha channels jointly. We design an effective variational autoencoder (VAE) that encodes the alpha channel into the RGB latent space. Then, to support the training of our diffusion transformer, we construct a high-quality and diverse RGBA video dataset. Compared with state-of-the-art methods, our model demonstrates superior performance in visual quality, motion realism, and transparency rendering. Notably, our model can generate a wide variety of semi-transparent objects, glowing effects, and fine-grained details such as hair strands. The released model is available on our website: https://donghaotian123.github.io/Wan-Alpha/.

Wenjing Wang, Wenhan Yang, Jiaying Liu

Face detection in low light scenarios is challenging but vital to many practical applications, e.g., surveillance video, autonomous driving at night. Most existing face detectors heavily rely on extensive annotations, while collecting data is time-consuming and laborious. To reduce the burden of building new datasets for low light conditions, we make full use of existing normal light data and explore how to adapt face detectors from normal light to low light. The challenge of this task is that the gap between normal and low light is too huge and complex for both pixel-level and object-level. Therefore, most existing low-light enhancement and adaptation methods do not achieve desirable performance. To address the issue, we propose a joint High-Low Adaptation (HLA) framework. Through a bidirectional low-level adaptation and multi-task high-level adaptation scheme, our HLA-Face outperforms state-of-the-art methods even without using dark face labels for training. Our project is publicly available at https://daooshee.github.io/HLA-Face-Website/

Yu Han, Shuai Yang, Wenjing Wang et al.

Fashion manipulation has attracted growing interest due to its great application value, which inspires many researches towards fashion images. However, little attention has been paid to fashion design draft. In this paper, we study a new unaligned translation problem between design drafts and real fashion items, whose main challenge lies in the huge misalignment between the two modalities. We first collect paired design drafts and real fashion item images without pixel-wise alignment. To solve the misalignment problem, our main idea is to train a sampling network to adaptively adjust the input to an intermediate state with structure alignment to the output. Moreover, built upon the sampling network, we present design draft to real fashion item translation network (D2RNet), where two separate translation streams that focus on texture and shape, respectively, are combined tactfully to get both benefits. D2RNet is able to generate realistic garments with both texture and shape consistency to their design drafts. We show that this idea can be effectively applied to the reverse translation problem and present R2DNet accordingly. Extensive experiments on unaligned fashion design translation demonstrate the superiority of our method over state-of-the-art methods. Our project website is available at: https://victoriahy.github.io/MM2020/ .

Shuai Yang, Wenjing Wang, Jiaying Liu

Text effects are combinations of visual elements such as outlines, colors and textures of text, which can dramatically improve its artistry. Although text effects are extensively utilized in the design industry, they are usually created by human experts due to their extreme complexity; this is laborious and not practical for normal users. In recent years, some efforts have been made toward automatic text effect transfer; however, the lack of data limits the capabilities of transfer models. To address this problem, we introduce a new text effects dataset, TE141K, with 141,081 text effect/glyph pairs in total. Our dataset consists of 152 professionally designed text effects rendered on glyphs, including English letters, Chinese characters, and Arabic numerals. To the best of our knowledge, this is the largest dataset for text effect transfer to date. Based on this dataset, we propose a baseline approach called text effect transfer GAN (TET-GAN), which supports the transfer of all 152 styles in one model and can efficiently extend to new styles. Finally, we conduct a comprehensive comparison in which 14 style transfer models are benchmarked. Experimental results demonstrate the superiority of TET-GAN both qualitatively and quantitatively and indicate that our dataset is effective and challenging.

Shuai Yang, Jiaying Liu, Wenjing Wang et al.

Text effects transfer technology automatically makes the text dramatically more impressive. However, previous style transfer methods either study the model for general style, which cannot handle the highly-structured text effects along the glyph, or require manual design of subtle matching criteria for text effects. In this paper, we focus on the use of the powerful representation abilities of deep neural features for text effects transfer. For this purpose, we propose a novel Texture Effects Transfer GAN (TET-GAN), which consists of a stylization subnetwork and a destylization subnetwork. The key idea is to train our network to accomplish both the objective of style transfer and style removal, so that it can learn to disentangle and recombine the content and style features of text effects images. To support the training of our network, we propose a new text effects dataset with as much as 64 professionally designed styles on 837 characters. We show that the disentangled feature representations enable us to transfer or remove all these styles on arbitrary glyphs using one network. Furthermore, the flexible network design empowers TET-GAN to efficiently extend to a new text style via one-shot learning where only one example is required. We demonstrate the superiority of the proposed method in generating high-quality stylized text over the state-of-the-art methods.

Chen Wei, Wenjing Wang, Wenhan Yang et al.

Retinex model is an effective tool for low-light image enhancement. It assumes that observed images can be decomposed into the reflectance and illumination. Most existing Retinex-based methods have carefully designed hand-crafted constraints and parameters for this highly ill-posed decomposition, which may be limited by model capacity when applied in various scenes. In this paper, we collect a LOw-Light dataset (LOL) containing low/normal-light image pairs and propose a deep Retinex-Net learned on this dataset, including a Decom-Net for decomposition and an Enhance-Net for illumination adjustment. In the training process for Decom-Net, there is no ground truth of decomposed reflectance and illumination. The network is learned with only key constraints including the consistent reflectance shared by paired low/normal-light images, and the smoothness of illumination. Based on the decomposition, subsequent lightness enhancement is conducted on illumination by an enhancement network called Enhance-Net, and for joint denoising there is a denoising operation on reflectance. The Retinex-Net is end-to-end trainable, so that the learned decomposition is by nature good for lightness adjustment. Extensive experiments demonstrate that our method not only achieves visually pleasing quality for low-light enhancement but also provides a good representation of image decomposition.