Wenqing Chu
Wenqing Chu has a total of 42 co-authors. Here are the top co-authors by number of shared publications:
- Deng Cai
State Key Lab of CAD&CG, Zhejiang University, Hangzhou, China
5 shared publications Deng Cai
State Key Laboratory of CAD&CG, Zhejiang University, Hangzhou, China
3 shared publications- Hongyang Xue
State Key Laboratory of CAD&CG, Zhejiang University, Hangzhou, China
3 shared publications - Chengjie Wang
YouTu Lab, Tencent, Shanghai, China
2 shared publications - Chengwei Yao
Zhejiang Provincial Key Laboratory of Service Robot, College of Computer Science, Zhejiang University, Hangzhou, China; Alibaba-Zhejiang University Joint Institute of Frontier Technologies, Hangzhou, China
2 shared publications - Boxi Wu
Zhejiang University
1 shared publication
Latest Publications
- Open Access
- 17 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 57288-57301
AVI-Talking: Learning Audio-Visual Instructions for Expressive 3D Talking Face Generation
While considerable progress has been made in achieving accurate lip synchronization for 3D speech-driven talking face generation, the task of incorporating expressive facial detail synthesis aligned with the speaker’s speaking status remains challenging. Existing efforts either focus on learning a dynamic talking head pose synchronized with speech rhythm or aim for stylized facial movements guided by external reference such as emotional labels or reference video clips. The former works often yield coarse alignment, neglecting the emotional nuances present in the audio content while the latter studies lead to unnatural applications, requiring manual style source selection by users. Our goal is to directly leverage the inherent style information conveyed by human speech for generating an expressive talking face that aligns with the speaking status. In this paper, we propose AVI-Talking, an Audio-Visual Instruction system for expressive Talking face generation. This system harnesses the robust contextual reasoning and hallucination capability offered by Large Language Models (LLMs) to instruct the realistic synthesis of 3D talking faces. Instead of directly learning facial movements from human speech, our two-stage strategy involves the LLMs first comprehending audio information and generating instructions implying expressive facial details seamlessly corresponding to the speech. Subsequently, a diffusion-based generative network executes these instructions. This two-stage process, coupled with the incorporation of LLMs, enhances model interpretability and provides users with flexibility to comprehend instructions and specify desired operations or modifications. Specifically, given a speech clip, we first employ a Q-Former for contrastive alignment the speech features with visual instructions, which is then projected to input text embedding of LLMs. It functions as a prompting strategy, prompting LLMs to generate plausible visual instructions that encompass diverse facial details. In order to use these predicted instructions, a language-guided talking face generation system with disentangled latent space is delicately derived, where the speech content related lip movements and emotion correlated facial expressions are separately represented in speech content space and content irrelevant space. Additionally, we introduce a contrastive instruction-style alignment and diffusion technique within the content-irrelevant space to fully exploit the talking prior network for diverse instruction-following synthesis. Extensive experiments showcase the effectiveness of our approach in producing vivid talking faces with expressive facial movements and consistent emotional status.
- 17 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 57288-57301
Open AccessAVI-Talking: Learning Audio-Visual Instructions for Expressive 3D Talking Face Generation
While considerable progress has been made in achieving accurate lip synchronization for 3D speech-driven talking face generation, the task of incorporating expressive facial detail synthesis aligned with the speaker’s speaking status remains challenging. Existing efforts either focus on learning a dynamic talking head pose synchronized with speech rhythm or aim for stylized facial movements guided by external reference such as emotional labels or reference video clips. The former works often yield coarse alignment, neglecting the emotional nuances present in the audio content while the latter studies lead to unnatural applications, requiring manual style source selection by users. Our goal is to directly leverage the inherent style information conveyed by human speech for generating an expressive talking face that aligns with the speaking status. In this paper, we propose AVI-Talking, an Audio-Visual Instruction system for expressive Talking face generation. This system harnesses the robust contextual reasoning and hallucination capability offered by Large Language Models (LLMs) to instruct the realistic synthesis of 3D talking faces. Instead of directly learning facial movements from human speech, our two-stage strategy involves the LLMs first comprehending audio information and generating instructions implying expressive facial details seamlessly corresponding to the speech. Subsequently, a diffusion-based generative network executes these instructions. This two-stage process, coupled with the incorporation of LLMs, enhances model interpretability and provides users with flexibility to comprehend instructions and specify desired operations or modifications. Specifically, given a speech clip, we first employ a Q-Former for contrastive alignment the speech features with visual instructions, which is then projected to input text embedding of LLMs. It functions as a prompting strategy, prompting LLMs to generate plausible visual instructions that encompass diverse facial details. In order to use these predicted instructions, a language-guided talking face generation system with disentangled latent space is delicately derived, where the speech content related lip movements and emotion correlated facial expressions are separately represented in speech content space and content irrelevant space. Additionally, we introduce a contrastive instruction-style alignment and diffusion technique within the content-irrelevant space to fully exploit the talking prior network for diverse instruction-following synthesis. Extensive experiments showcase the effectiveness of our approach in producing vivid talking faces with expressive facial movements and consistent emotional status.
- 19 September 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 32 (10577149) , 5296-5309
Dynamic Frame Interpolation in Wavelet Domain
Video frame interpolation is an important low-level vision task, which can increase frame rate for more fluent visual experience. Existing methods have achieved great success by employing advanced motion models and synthesis networks. However, the spatial redundancy when synthesizing the target frame has not been fully explored, that can result in lots of inefficient computation. On the other hand, the computation compression degree in frame interpolation is highly dependent on both texture distribution and scene motion, which demands to understand the spatial-temporal information of each input frame pair for a better compression degree selection. In this work, we propose a novel two-stage frame interpolation framework termed WaveletVFI to address above problems. It first estimates intermediate optical flow with a lightweight motion perception network, and then a wavelet synthesis network uses flow aligned context features to predict multi-scale wavelet coefficients with sparse convolution for efficient target frame reconstruction, where the sparse valid masks that control computation in each scale are determined by a crucial threshold ratio. Instead of setting a fixed value like previous methods, we find that embedding a classifier in the motion perception network to learn a dynamic threshold for each sample can achieve more computation reduction with almost no loss of accuracy. On the common high resolution and animation frame interpolation benchmarks, proposed WaveletVFI can reduce computation up to 40% while maintaining similar accuracy, making it perform more efficiently against other state-of-the-arts.
- 19 September 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 32 (10577149) , 5296-5309
Dynamic Frame Interpolation in Wavelet Domain
Video frame interpolation is an important low-level vision task, which can increase frame rate for more fluent visual experience. Existing methods have achieved great success by employing advanced motion models and synthesis networks. However, the spatial redundancy when synthesizing the target frame has not been fully explored, that can result in lots of inefficient computation. On the other hand, the computation compression degree in frame interpolation is highly dependent on both texture distribution and scene motion, which demands to understand the spatial-temporal information of each input frame pair for a better compression degree selection. In this work, we propose a novel two-stage frame interpolation framework termed WaveletVFI to address above problems. It first estimates intermediate optical flow with a lightweight motion perception network, and then a wavelet synthesis network uses flow aligned context features to predict multi-scale wavelet coefficients with sparse convolution for efficient target frame reconstruction, where the sparse valid masks that control computation in each scale are determined by a crucial threshold ratio. Instead of setting a fixed value like previous methods, we find that embedding a classifier in the motion perception network to learn a dynamic threshold for each sample can achieve more computation reduction with almost no loss of accuracy. On the common high resolution and animation frame interpolation benchmarks, proposed WaveletVFI can reduce computation up to 40% while maintaining similar accuracy, making it perform more efficiently against other state-of-the-arts.
- Open Access
High-fidelity Generalized Emotional Talking Face Generation with Multi-modal Emotion Space Learning
Recently, emotional talking face generation has received considerable attention. However, existing methods only adopt one-hot coding, image, or audio as emotion conditions, thus lacking flexible control in practical applications and failing to handle unseen emotion styles due to limited semantics. They either ignore the one-shot setting or the quality of generated faces. In this paper, we propose a more flexible and generalized framework. Specifically, we supplement the emotion style in text prompts and use an Aligned Multi-modal Emotion encoder to embed the text, image, and audio emotion modality into a unified space, which inherits rich semantic prior from CLIP. Consequently, effective multi-modal emotion space learning helps our method support arbitrary emotion modality during testing and could generalize to unseen emotion styles. Besides, an Emotion-aware Audio-to-3DMM Convertor is proposed to connect the emotion condition and the audio sequence to structural representation. A followed style-based High-fidelity Emotional Face generator is designed to generate arbitrary high-resolution realistic identities. Our texture generator hierarchically learns flow fields and animated faces in a residual manner. Extensive experiments demonstrate the flexibility and generalization of our method in emotion control and the effectiveness of high-quality face synthesis.
Open AccessHigh-fidelity Generalized Emotional Talking Face Generation with Multi-modal Emotion Space Learning
Recently, emotional talking face generation has received considerable attention. However, existing methods only adopt one-hot coding, image, or audio as emotion conditions, thus lacking flexible control in practical applications and failing to handle unseen emotion styles due to limited semantics. They either ignore the one-shot setting or the quality of generated faces. In this paper, we propose a more flexible and generalized framework. Specifically, we supplement the emotion style in text prompts and use an Aligned Multi-modal Emotion encoder to embed the text, image, and audio emotion modality into a unified space, which inherits rich semantic prior from CLIP. Consequently, effective multi-modal emotion space learning helps our method support arbitrary emotion modality during testing and could generalize to unseen emotion styles. Besides, an Emotion-aware Audio-to-3DMM Convertor is proposed to connect the emotion condition and the audio sequence to structural representation. A followed style-based High-fidelity Emotional Face generator is designed to generate arbitrary high-resolution realistic identities. Our texture generator hierarchically learns flow fields and animated faces in a residual manner. Extensive experiments demonstrate the flexibility and generalization of our method in emotion control and the effectiveness of high-quality face synthesis.
- Open Access
CFNet: Learning Correlation Functions for One-Stage Panoptic Segmentation
Recently, there is growing attention on one-stage panoptic segmentation methods which aim to segment instances and stuff jointly within a fully convolutional pipeline efficiently. However, most of the existing works directly feed the backbone features to various segmentation heads ignoring the demands for semantic and instance segmentation are different: The former needs semantic-level discriminative features, while the latter requires features to be distinguishable across instances. To alleviate this, we propose to first predict semantic-level and instance-level correlations among different locations that are utilized to enhance the backbone features, and then feed the improved discriminative features into the corresponding segmentation heads, respectively. Specifically, we organize the correlations between a given location and all locations as a continuous sequence and predict it as a whole. Considering that such a sequence can be extremely complicated, we adopt Discrete Fourier Transform (DFT), a tool that can approximate an arbitrary sequence parameterized by amplitudes and phrases. For different tasks, we generate these parameters from the backbone features in a fully convolutional way which is optimized implicitly by corresponding tasks. As a result, these accurate and consistent correlations contribute to producing plausible discriminative features which meet the requirements of the complicated panoptic segmentation task. To verify the effectiveness of our methods, we conduct experiments on several challenging panoptic segmentation datasets and achieve state-of-the-art performance on MS COCO with $45.1$\% PQ and ADE20k with $32.6$\% PQ.
Open AccessCFNet: Learning Correlation Functions for One-Stage Panoptic Segmentation
Recently, there is growing attention on one-stage panoptic segmentation methods which aim to segment instances and stuff jointly within a fully convolutional pipeline efficiently. However, most of the existing works directly feed the backbone features to various segmentation heads ignoring the demands for semantic and instance segmentation are different: The former needs semantic-level discriminative features, while the latter requires features to be distinguishable across instances. To alleviate this, we propose to first predict semantic-level and instance-level correlations among different locations that are utilized to enhance the backbone features, and then feed the improved discriminative features into the corresponding segmentation heads, respectively. Specifically, we organize the correlations between a given location and all locations as a continuous sequence and predict it as a whole. Considering that such a sequence can be extremely complicated, we adopt Discrete Fourier Transform (DFT), a tool that can approximate an arbitrary sequence parameterized by amplitudes and phrases. For different tasks, we generate these parameters from the backbone features in a fully convolutional way which is optimized implicitly by corresponding tasks. As a result, these accurate and consistent correlations contribute to producing plausible discriminative features which meet the requirements of the complicated panoptic segmentation task. To verify the effectiveness of our methods, we conduct experiments on several challenging panoptic segmentation datasets and achieve state-of-the-art performance on MS COCO with $45.1$\% PQ and ADE20k with $32.6$\% PQ.
- Open Access
- 9 July 2021
- Published by Springer Nature in International Journal of Computer Vision
- Vol. 129 (9) , 2663-2679
Learning to Caricature via Semantic Shape Transform
Caricature is an artistic drawing created to abstract or exaggerate facial features of a person. Rendering visually pleasing caricatures is a difficult task that requires professional skills, and thus it is of great interest to design a method to automatically generate such drawings. To deal with large shape changes, we propose an algorithm based on a semantic shape transform to produce diverse and plausible shape exaggerations. Specifically, we predict pixel-wise semantic correspondences and perform image warping on the input photo to achieve dense shape transformation. We show that the proposed framework is able to render visually pleasing shape exaggerations while maintaining their facial structures. In addition, our model allows users to manipulate the shape via the semantic map. We demonstrate the effectiveness of our approach on a large photograph-caricature benchmark dataset with comparisons to the state-of-the-art methods.
- 9 July 2021
- Published by Springer Nature in International Journal of Computer Vision
- Vol. 129 (9) , 2663-2679
Open AccessLearning to Caricature via Semantic Shape Transform
Caricature is an artistic drawing created to abstract or exaggerate facial features of a person. Rendering visually pleasing caricatures is a difficult task that requires professional skills, and thus it is of great interest to design a method to automatically generate such drawings. To deal with large shape changes, we propose an algorithm based on a semantic shape transform to produce diverse and plausible shape exaggerations. Specifically, we predict pixel-wise semantic correspondences and perform image warping on the input photo to achieve dense shape transformation. We show that the proposed framework is able to render visually pleasing shape exaggerations while maintaining their facial structures. In addition, our model allows users to manipulate the shape via the semantic map. We demonstrate the effectiveness of our approach on a large photograph-caricature benchmark dataset with comparisons to the state-of-the-art methods.
- Open Access
Improving Semantic Segmentation via Dilated Affinity
Introducing explicit constraints on the structural predictions has been an effective way to improve the performance of semantic segmentation models. Existing methods are mainly based on insufficient hand-crafted rules that only partially capture the image structure, and some methods can also suffer from the efficiency issue. As a result, most of the state-of-the-art fully convolutional networks did not adopt these techniques. In this work, we propose a simple, fast yet effective method that exploits structural information through direct supervision with minor additional expense. To be specific, our method explicitly requires the network to predict semantic segmentation as well as dilated affinity, which is a sparse version of pair-wise pixel affinity. The capability of telling the relationships between pixels are directly built into the model and enhance the quality of segmentation in two stages. 1) Joint training with dilated affinity can provide robust feature representations and thus lead to finer segmentation results. 2) The extra output of affinity information can be further utilized to refine the original segmentation with a fast propagation process. Consistent improvements are observed on various benchmark datasets when applying our framework to the existing state-of-the-art model. Codes will be released soon.
Open AccessImproving Semantic Segmentation via Dilated Affinity
Introducing explicit constraints on the structural predictions has been an effective way to improve the performance of semantic segmentation models. Existing methods are mainly based on insufficient hand-crafted rules that only partially capture the image structure, and some methods can also suffer from the efficiency issue. As a result, most of the state-of-the-art fully convolutional networks did not adopt these techniques. In this work, we propose a simple, fast yet effective method that exploits structural information through direct supervision with minor additional expense. To be specific, our method explicitly requires the network to predict semantic segmentation as well as dilated affinity, which is a sparse version of pair-wise pixel affinity. The capability of telling the relationships between pixels are directly built into the model and enhance the quality of segmentation in two stages. 1) Joint training with dilated affinity can provide robust feature representations and thus lead to finer segmentation results. 2) The extra output of affinity information can be further utilized to refine the original segmentation with a fast propagation process. Consistent improvements are observed on various benchmark datasets when applying our framework to the existing state-of-the-art model. Codes will be released soon.
- 25 July 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 27 (11) , 5563-5574
A Better Way to Attend: Attention With Trees for Video Question Answering
We propose a new attention model for video question answering. The main idea of the attention models is to locate on the most informative parts of the visual data. The attention mechanisms are quite popular these days. However, most existing visual attention mechanisms regard the question as a whole. They ignore the word-level semantics where each word can have different attentions and some words need no attention. Neither do they consider the semantic structure of the sentences. Although the extended soft attention model for video question answering leverages the word-level attention, it performs poorly on long question sentences. In this paper, we propose the heterogeneous tree-structured memory network (HTreeMN) for video question answering. Our proposed approach is based upon the syntax parse trees of the question sentences. The HTreeMN treats the words differently where the visual words are processed with an attention module and the verbal ones not. It also utilizes the semantic structure of the sentences by combining the neighbors based on the recursive structure of the parse trees. The understandings of the words and the videos are propagated and merged from leaves to the root. Furthermore, we build a hierarchical attention mechanism to distill the attended features. We evaluate our approach on two data sets. The experimental results show the superiority of our HTreeMN model over the other attention models, especially on complex questions.
- 25 July 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 27 (11) , 5563-5574
A Better Way to Attend: Attention With Trees for Video Question Answering
We propose a new attention model for video question answering. The main idea of the attention models is to locate on the most informative parts of the visual data. The attention mechanisms are quite popular these days. However, most existing visual attention mechanisms regard the question as a whole. They ignore the word-level semantics where each word can have different attentions and some words need no attention. Neither do they consider the semantic structure of the sentences. Although the extended soft attention model for video question answering leverages the word-level attention, it performs poorly on long question sentences. In this paper, we propose the heterogeneous tree-structured memory network (HTreeMN) for video question answering. Our proposed approach is based upon the syntax parse trees of the question sentences. The HTreeMN treats the words differently where the visual words are processed with an attention module and the verbal ones not. It also utilizes the semantic structure of the sentences by combining the neighbors based on the recursive structure of the parse trees. The understandings of the words and the videos are propagated and merged from leaves to the root. Furthermore, we build a hierarchical attention mechanism to distill the attended features. We evaluate our approach on two data sets. The experimental results show the superiority of our HTreeMN model over the other attention models, especially on complex questions.
- 7 July 2018
- Published by Elsevier in Neurocomputing
- Vol. 314, 386-393
The forgettable-watcher model for video question answering
- 7 July 2018
- Published by Elsevier in Neurocomputing
- Vol. 314, 386-393
The forgettable-watcher model for video question answering
- 11 June 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Multimedia
- Vol. 21 (1) , 246-255
Sparse Coding Guided Spatiotemporal Feature Learning for Abnormal Event Detection in Large Videos
Abnormal event detection in large videos is an important task in research and industrial applications, which has attracted considerable attention in recent years. Existing methods usually solve this problem by extracting local features and then learning an outlier detection model on training videos. However, most previous approaches merely employ hand-crafted visual features, which is a clear disadvantage due to their limited representation capacity. In this paper, we present a novel unsupervised deep feature learning algorithm for the abnormal event detection problem. To exploit the spatiotemporal information of the inputs, we utilize the deep 3-dimensional convolutional network (C3D) to perform feature extraction. Then the key problem is how to train the C3D network without any category labels. Here, we employ the sparse coding results of the hand-crafted features generated from the inputs to guide the unsupervised feature learning. Specifically, we define a multi-level similarity relationship between these inputs according to the statistical information of the shared atoms. In the following, we introduce the quadruplet concept to model the multi-level similarity structure, which could be used to construct a generalized triplet loss for training the C3D network. Furthermore, the C3D network could be utilized to generate the features for sparse coding again, and this pipeline could be iterated for several times. By jointly optimizing between the sparse coding and the unsupervised feature learning, we can obtain robust and rich feature representations. Based on the learned representations, the sparse reconstruction error is applied to predicting the anomaly score of each testing input. Experiments on several publicly available video surveillance datasets in comparison with a number of existing works demonstrate that the proposed algorithm performs favorably against the state-of-the-art methods.
- 11 June 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Multimedia
- Vol. 21 (1) , 246-255
Sparse Coding Guided Spatiotemporal Feature Learning for Abnormal Event Detection in Large Videos
Abnormal event detection in large videos is an important task in research and industrial applications, which has attracted considerable attention in recent years. Existing methods usually solve this problem by extracting local features and then learning an outlier detection model on training videos. However, most previous approaches merely employ hand-crafted visual features, which is a clear disadvantage due to their limited representation capacity. In this paper, we present a novel unsupervised deep feature learning algorithm for the abnormal event detection problem. To exploit the spatiotemporal information of the inputs, we utilize the deep 3-dimensional convolutional network (C3D) to perform feature extraction. Then the key problem is how to train the C3D network without any category labels. Here, we employ the sparse coding results of the hand-crafted features generated from the inputs to guide the unsupervised feature learning. Specifically, we define a multi-level similarity relationship between these inputs according to the statistical information of the shared atoms. In the following, we introduce the quadruplet concept to model the multi-level similarity structure, which could be used to construct a generalized triplet loss for training the C3D network. Furthermore, the C3D network could be utilized to generate the features for sparse coding again, and this pipeline could be iterated for several times. By jointly optimizing between the sparse coding and the unsupervised feature learning, we can obtain robust and rich feature representations. Based on the learned representations, the sparse reconstruction error is applied to predicting the anomaly score of each testing input. Experiments on several publicly available video surveillance datasets in comparison with a number of existing works demonstrate that the proposed algorithm performs favorably against the state-of-the-art methods.
- 1 January 2018
- Published by Elsevier in Neurocomputing
- Vol. 275, 1035-1042
Deep feature based contextual model for object detection
- 1 January 2018
- Published by Elsevier in Neurocomputing
- Vol. 275, 1035-1042
Deep feature based contextual model for object detection