Zeng Zhao
Research Subjects
Zeng Zhao has a total of 43 co-authors. Here are the top co-authors by number of shared publications:
- Zhipeng Hu
Fuxi AI Lab, NetEase Inc., Hangzhou, China
4 shared publications - Changjie Fan
NetEase Fuxi AI Lab, Hangzhou, China
3 shared publications - Rongsheng Zhang
Fuxi AI Lab Netease Inc
3 shared publications - Tangjie Lv
NetEase Fuxi AI Lab, Hangzhou, China
2 shared publications - Yu Ding
NetEase Fuxi AI Lab
2 shared publications - Yuhang Ma
Fuxi AI Lab, NetEase
2 shared publications
Latest Publications
- Top Cited PapersOpen Access
- 11 April 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 39 (5) , 5523-5531
LLM4GEN: Leveraging Semantic Representation of LLMs for Text-to-Image Generation
Diffusion models have exhibited substantial success in text-to-image generation. However, they often encounter challenges when dealing with complex and dense prompts involving multiple objects, attribute binding, and long descriptions. In this paper, we propose a novel framework called LLM4GEN, which enhances the semantic understanding of text-to-image diffusion models by leveraging the representation of Large Language Models (LLMs). It can be seamlessly incorporated into various diffusion models as a plug-and-play component. A specially designed Cross-Adapter Module (CAM) integrates the original text features of text-to-image models with LLM features, thereby enhancing text-to-image generation. Additionally, to facilitate and correct entity-attribute relationships in text prompts, we develop an entity-guided regularization loss to further improve generation performance. We also introduce DensePrompts, which contains 7,000 dense prompts to provide a comprehensive evaluation for the text-to-image generation task. Experiments indicate that LLM4GEN significantly improves the semantic alignment of SD1.5 and SDXL, demonstrating increases of 9.69% and 12.90% in color on T2I-CompBench, respectively. Moreover, it surpasses existing models in terms of sample quality, image-text alignment, and human evaluation.
- 11 April 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 39 (5) , 5523-5531
Top Cited PapersOpen AccessLLM4GEN: Leveraging Semantic Representation of LLMs for Text-to-Image Generation
Diffusion models have exhibited substantial success in text-to-image generation. However, they often encounter challenges when dealing with complex and dense prompts involving multiple objects, attribute binding, and long descriptions. In this paper, we propose a novel framework called LLM4GEN, which enhances the semantic understanding of text-to-image diffusion models by leveraging the representation of Large Language Models (LLMs). It can be seamlessly incorporated into various diffusion models as a plug-and-play component. A specially designed Cross-Adapter Module (CAM) integrates the original text features of text-to-image models with LLM features, thereby enhancing text-to-image generation. Additionally, to facilitate and correct entity-attribute relationships in text prompts, we develop an entity-guided regularization loss to further improve generation performance. We also introduce DensePrompts, which contains 7,000 dense prompts to provide a comprehensive evaluation for the text-to-image generation task. Experiments indicate that LLM4GEN significantly improves the semantic alignment of SD1.5 and SDXL, demonstrating increases of 9.69% and 12.90% in color on T2I-CompBench, respectively. Moreover, it surpasses existing models in terms of sample quality, image-text alignment, and human evaluation.
- Open Access
- 11 April 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 39 (6) , 6027-6035
Storynizor: Consistent Story Generation via Inter-Frame Synchronized and Shuffled ID Injection
Recent advances in text-to-image diffusion models have spurred significant interest in continuous story image generation. In this paper, we introduce Storynizor, a model capable of generating coherent stories with strong inter-frame character consistency, effective foreground-background separation, and diverse pose variation. The core innovation of Storynizor lies in its key modules: ID-Synchronizer and ID-Injector. The ID-Synchronizer employs an auto-mask self-attention module and a mask perceptual loss across inter-frame images to improve the consistency of character generation, vividly representing their postures and backgrounds. The ID-Injector utilize a Shuffling Reference Strategy (SRS) to integrate ID features into specific locations, enhancing ID-based consistent character generation. Additionally, to facilitate the training of Storynizor, we have curated a novel dataset called StoryDB comprising 100, 000 images. This dataset contains single and multiple-character sets in diverse environments, layouts, and gestures with detailed descriptions. Experimental results indicate that Storynizor demonstrates superior coherent story generation with high-fidelity character consistency, flexible postures, and vivid backgrounds compared to other character-specific methods.
- 11 April 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 39 (6) , 6027-6035
Open AccessStorynizor: Consistent Story Generation via Inter-Frame Synchronized and Shuffled ID Injection
Recent advances in text-to-image diffusion models have spurred significant interest in continuous story image generation. In this paper, we introduce Storynizor, a model capable of generating coherent stories with strong inter-frame character consistency, effective foreground-background separation, and diverse pose variation. The core innovation of Storynizor lies in its key modules: ID-Synchronizer and ID-Injector. The ID-Synchronizer employs an auto-mask self-attention module and a mask perceptual loss across inter-frame images to improve the consistency of character generation, vividly representing their postures and backgrounds. The ID-Injector utilize a Shuffling Reference Strategy (SRS) to integrate ID features into specific locations, enhancing ID-based consistent character generation. Additionally, to facilitate the training of Storynizor, we have curated a novel dataset called StoryDB comprising 100, 000 images. This dataset contains single and multiple-character sets in diverse environments, layouts, and gestures with detailed descriptions. Experimental results indicate that Storynizor demonstrates superior coherent story generation with high-fidelity character consistency, flexible postures, and vivid backgrounds compared to other character-specific methods.
- 5 January 2025
- Published by Elsevier in Neurocomputing
- Vol. 622
Let storytelling tell vivid stories: A multi-modal-agent-based unified storytelling framework
- 5 January 2025
- Published by Elsevier in Neurocomputing
- Vol. 622
Let storytelling tell vivid stories: A multi-modal-agent-based unified storytelling framework
- Open Access
- 28 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 5653-5662
Selection and Reconstruction of Key Locals: A Novel Specific Domain Image-Text Retrieval Method
In recent years, Vision-Language Pre-training (VLP) models have demonstrated rich prior knowledge for multimodal alignment, prompting investigations into their application in Specific Domain Image-Text Retrieval(SDITR) such as Text-Image Person Re-identification (TIReID) and Remote Sensing Image-Text Retrieval (RSITR). Due to the unique data characteristics in specific scenarios, the primary challenge is to leverage discriminative fine-grained local information for improved mapping of images and text into a shared space. Current approaches interact with all multimodal local features for alignment, implicitly focusing on discriminative local information to distinguish data differences, which may bring noise and uncertainty. Furthermore, their VLP feature extractors like CLIP often focus on instance-level representations, potentially reducing the discriminability of fine-grained local features. To alleviate these issues, we propose an Explicit Key Local information Selection and Reconstruction Framework (EKLSR), which explicitly selects key local information to enhance feature representation. Specifically, we introduce a Key Local information Selection and Fusion (KLSF) that utilizes hidden knowledge from the VLP model to select interpretably and fuse key local information. Secondly, we employ Key Local segment Reconstruction (KLR) based on multimodal interaction to reconstruct the key local segments of images (text), significantly enriching their discriminative information and enhancing both inter-modal and intra-modal interaction alignment. To demonstrate the effectiveness of our approach, we conducted experiments on five datasets across TIReID and RSITR. Notably, our EKLSR model achieves state-of-the-art performance on two RSITR datasets.
- 28 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 5653-5662
Open AccessSelection and Reconstruction of Key Locals: A Novel Specific Domain Image-Text Retrieval Method
In recent years, Vision-Language Pre-training (VLP) models have demonstrated rich prior knowledge for multimodal alignment, prompting investigations into their application in Specific Domain Image-Text Retrieval(SDITR) such as Text-Image Person Re-identification (TIReID) and Remote Sensing Image-Text Retrieval (RSITR). Due to the unique data characteristics in specific scenarios, the primary challenge is to leverage discriminative fine-grained local information for improved mapping of images and text into a shared space. Current approaches interact with all multimodal local features for alignment, implicitly focusing on discriminative local information to distinguish data differences, which may bring noise and uncertainty. Furthermore, their VLP feature extractors like CLIP often focus on instance-level representations, potentially reducing the discriminability of fine-grained local features. To alleviate these issues, we propose an Explicit Key Local information Selection and Reconstruction Framework (EKLSR), which explicitly selects key local information to enhance feature representation. Specifically, we introduce a Key Local information Selection and Fusion (KLSF) that utilizes hidden knowledge from the VLP model to select interpretably and fuse key local information. Secondly, we employ Key Local segment Reconstruction (KLR) based on multimodal interaction to reconstruct the key local segments of images (text), significantly enriching their discriminative information and enhancing both inter-modal and intra-modal interaction alignment. To demonstrate the effectiveness of our approach, we conducted experiments on five datasets across TIReID and RSITR. Notably, our EKLSR model achieves state-of-the-art performance on two RSITR datasets.
- 19 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Affective Computing
- Vol. 15 (3) , 1669-1683
Facial Action Unit Detection and Intensity Estimation From Self-Supervised Representation
As a fine-grained and local expression behavior measurement, facial action unit (FAU) analysis (e.g., detection and intensity estimation) has been documented for its time-consuming, labor-intensive, and error-prone annotation. Thus a long-standing challenge of FAU analysis arises from the data scarcity of manual annotations, limiting the generalization ability of trained models to a large extent. Amounts of previous works have made efforts to alleviate this issue via semi/weakly supervised methods and extra auxiliary information. However, these methods still require domain knowledge and have not yet avoided the high dependency on data annotation. This paper introduces a robust facial representation model MAE-Face for AU analysis. Using masked autoencoding as the self-supervised pre-training approach, MAE-Face first learns a high-capacity model from a feasible collection of face images without additional data annotations. Then after being fine-tuned on AU datasets, MAE-Face exhibits convincing performance for both AU detection and AU intensity estimation, achieving a new state-of-the-art on nearly all the evaluation results. Further investigation shows that MAE-Face achieves decent performance even when fine-tuned on only 1% of the AU training set, strongly proving its robustness and generalization performance. The pre-trained model is available at GitHub repository.
- 19 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Affective Computing
- Vol. 15 (3) , 1669-1683
Facial Action Unit Detection and Intensity Estimation From Self-Supervised Representation
As a fine-grained and local expression behavior measurement, facial action unit (FAU) analysis (e.g., detection and intensity estimation) has been documented for its time-consuming, labor-intensive, and error-prone annotation. Thus a long-standing challenge of FAU analysis arises from the data scarcity of manual annotations, limiting the generalization ability of trained models to a large extent. Amounts of previous works have made efforts to alleviate this issue via semi/weakly supervised methods and extra auxiliary information. However, these methods still require domain knowledge and have not yet avoided the high dependency on data annotation. This paper introduces a robust facial representation model MAE-Face for AU analysis. Using masked autoencoding as the self-supervised pre-training approach, MAE-Face first learns a high-capacity model from a feasible collection of face images without additional data annotations. Then after being fine-tuned on AU datasets, MAE-Face exhibits convincing performance for both AU detection and AU intensity estimation, achieving a new state-of-the-art on nearly all the evaluation results. Further investigation shows that MAE-Face achieves decent performance even when fine-tuned on only 1% of the AU training set, strongly proving its robustness and generalization performance. The pre-trained model is available at GitHub repository.
- Open Access
- 26 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 3403-3414
Beyond First Impressions: Integrating Joint Multi-modal Cues for Comprehensive 3D Representation
In recent years, 3D representation learning has turned to 2D vision-language pre-trained models to overcome data scarcity challenges. However, existing methods simply transfer 2D alignment strategies, aligning 3D representations with single-view 2D images and coarse-grained parent category text. These approaches introduce information degradation and insufficient synergy issues, leading to performance loss. Information degradation arises from overlooking the fact that a 3D representation should be equivalent to a series of multi-view images and more fine-grained subcategory text. Insufficient synergy neglects the idea that a robust 3D representation should align with the joint vision-language space, rather than independently aligning with each modality. In this paper, we propose a multi-view joint modality modeling approach, termed JM3D, to obtain a unified representation for point cloud, text, and image. Specifically, a novel Structured Multimodal Organizer (SMO) is proposed to address the information degradation issue, which introduces contiguous multi-view images and hierarchical text to enrich the representation of vision and language modalities. A Joint Multi-modal Alignment (JMA) is designed to tackle the insufficient synergy problem, which models the joint modality by incorporating language knowledge into the visual modality. Extensive experiments on ModelNet40 and ScanObjectNN demonstrate the effectiveness of our proposed method, JM3D, which achieves state-of-the-art performance in zero-shot 3D classification. JM3D outperforms ULIP by approximately 4.3% on PointMLP and achieves an improvement of up to 6.5% accuracy on PointNet++ in top-1 accuracy for zero-shot 3D classification on ModelNet40. The source code and trained models for all our experiments are publicly available at https://github.com/Mr-Neko/JM3D.
- 26 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 3403-3414
Open AccessBeyond First Impressions: Integrating Joint Multi-modal Cues for Comprehensive 3D Representation
In recent years, 3D representation learning has turned to 2D vision-language pre-trained models to overcome data scarcity challenges. However, existing methods simply transfer 2D alignment strategies, aligning 3D representations with single-view 2D images and coarse-grained parent category text. These approaches introduce information degradation and insufficient synergy issues, leading to performance loss. Information degradation arises from overlooking the fact that a 3D representation should be equivalent to a series of multi-view images and more fine-grained subcategory text. Insufficient synergy neglects the idea that a robust 3D representation should align with the joint vision-language space, rather than independently aligning with each modality. In this paper, we propose a multi-view joint modality modeling approach, termed JM3D, to obtain a unified representation for point cloud, text, and image. Specifically, a novel Structured Multimodal Organizer (SMO) is proposed to address the information degradation issue, which introduces contiguous multi-view images and hierarchical text to enrich the representation of vision and language modalities. A Joint Multi-modal Alignment (JMA) is designed to tackle the insufficient synergy problem, which models the joint modality by incorporating language knowledge into the visual modality. Extensive experiments on ModelNet40 and ScanObjectNN demonstrate the effectiveness of our proposed method, JM3D, which achieves state-of-the-art performance in zero-shot 3D classification. JM3D outperforms ULIP by approximately 4.3% on PointMLP and achieves an improvement of up to 6.5% accuracy on PointNet++ in top-1 accuracy for zero-shot 3D classification on ModelNet40. The source code and trained models for all our experiments are publicly available at https://github.com/Mr-Neko/JM3D.
- 28 September 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Visualization and Computer Graphics
- Vol. 29 (2) , 1400-1414
A Music-Driven Deep Generative Adversarial Model for Guzheng Playing Animation
To date relatively few efforts have been made on the automatic generation of musical instrument playing animations. This problem is challenging due to the intrinsically complex, temporal relationship between music and human motion as well as the lacking of high quality music-playing motion datasets. In this article, we propose a fully automatic, deep learning based framework to synthesize realistic upper body animations based on novel guzheng music input. Specifically, based on a recorded audiovisual motion capture dataset, we delicately design a generative adversarial network (GAN) based approach to capture the temporal relationship between the music and the human motion data. In this process, data augmentation is employed to improve the generalization of our approach to handle a variety of guzheng music inputs. Through extensive objective and subjective experiments, we show that our method can generate visually plausible guzheng-playing animations that are well synchronized with the input guzheng music, and it can significantly outperform the state-of-the-art methods. In addition, through an ablation study, we validate the contributions of the carefully-designed modules in our framework.
- 28 September 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Visualization and Computer Graphics
- Vol. 29 (2) , 1400-1414
A Music-Driven Deep Generative Adversarial Model for Guzheng Playing Animation
To date relatively few efforts have been made on the automatic generation of musical instrument playing animations. This problem is challenging due to the intrinsically complex, temporal relationship between music and human motion as well as the lacking of high quality music-playing motion datasets. In this article, we propose a fully automatic, deep learning based framework to synthesize realistic upper body animations based on novel guzheng music input. Specifically, based on a recorded audiovisual motion capture dataset, we delicately design a generative adversarial network (GAN) based approach to capture the temporal relationship between the music and the human motion data. In this process, data augmentation is employed to improve the generalization of our approach to handle a variety of guzheng music inputs. Through extensive objective and subjective experiments, we show that our method can generate visually plausible guzheng-playing animations that are well synchronized with the input guzheng music, and it can significantly outperform the state-of-the-art methods. In addition, through an ablation study, we validate the contributions of the carefully-designed modules in our framework.