STAR: Scale-wise Text-to-image generation via Auto-Regressive representations

We present STAR, a text-to-image model that employs scale-wise auto-regressive paradigm. Unlike VAR, which is limited to class-conditioned synthesis within a fixed set of predetermined categories, our STAR enables text-driven open-set generation through three key designs: To boost diversity and generalizability with unseen combinations of objects and concepts, we introduce a pre-trained text encoder to extract representations for textual constraints, which we then use as guidance. To improve the interactions between generated images and fine-grained textual guidance, making results more controllable, additional cross-attention layers are incorporated at each scale. Given the natural structure correlation across different scales, we leverage 2D Rotary Positional Encoding (RoPE) and tweak it into a normalized version. This ensures consistent interpretation of relative positions across token maps at different scales and stabilizes the training process. Extensive experiments demonstrate that STAR surpasses existing benchmarks in terms of fidelity,image text consistency, and aesthetic quality. Our findings emphasize the potential of auto-regressive methods in the field of high-quality image synthesis, offering promising new directions for the T2I field currently dominated by diffusion methods.

Dynamic Prompt Optimizing for Text-to-Image Generation

Text-to-image generative models, specifically those based on diffusion models like Imagen and Stable Diffusion, have made substantial advancements. Recently, there has been a surge of interest in the delicate refinement of text prompts. Users assign weights or alter the injection time steps of certain words in the text prompts to improve the quality of generated images. However, the success of fine-control prompts depends on the accuracy of the text prompts and the careful selection of weights and time steps, which requires significant manual intervention. To address this, we introduce the \textbf{P}rompt \textbf{A}uto-\textbf{E}diting (PAE) method. Besides refining the original prompts for image generation, we further employ an online reinforcement learning strategy to explore the weights and injection time steps of each word, leading to the dynamic fine-control prompts. The reward function during training encourages the model to consider aesthetic score, semantic consistency, and user preferences. Experimental results demonstrate that our proposed method effectively improves the original prompts, generating visually more appealing images while maintaining semantic alignment. Code is available at https://github.com/Mowenyii/PAE.

StyleInject: Parameter Efficient Tuning of Text-to-Image Diffusion Models

The ability to fine-tune generative models for text-to-image generation tasks is crucial, particularly facing the complexity involved in accurately interpreting and visualizing textual inputs. While LoRA is efficient for language model adaptation, it often falls short in text-to-image tasks due to the intricate demands of image generation, such as accommodating a broad spectrum of styles and nuances. To bridge this gap, we introduce StyleInject, a specialized fine-tuning approach tailored for text-to-image models. StyleInject comprises multiple parallel low-rank parameter matrices, maintaining the diversity of visual features. It dynamically adapts to varying styles by adjusting the variance of visual features based on the characteristics of the input signal. This approach significantly minimizes the impact on the original model's text-image alignment capabilities while adeptly adapting to various styles in transfer learning. StyleInject proves particularly effective in learning from and enhancing a range of advanced, community-fine-tuned generative models. Our comprehensive experiments, including both small-sample and large-scale data fine-tuning as well as base model distillation, show that StyleInject surpasses traditional LoRA in both text-image semantic consistency and human preference evaluation, all while ensuring greater parameter efficiency.

Learning and Evaluating Human Preferences for Conversational Head Generation

A reliable and comprehensive evaluation metric that aligns with manual preference assessments is crucial for conversational head video synthesis methods development. Existing quantitative evaluations often fail to capture the full complexity of human preference, as they only consider limited evaluation dimensions. Qualitative evaluations and user studies offer a solution but are time-consuming and labor-intensive. This limitation hinders the advancement of conversational head generation algorithms and systems. In this paper, we propose a novel learning-based evaluation metric named Preference Score (PS) for fitting human preference according to the quantitative evaluations across different dimensions. PS can serve as a quantitative evaluation without the need for human annotation. Experimental results validate the superiority of Preference Score in aligning with human perception, and also demonstrate robustness and generalizability to unseen data, making it a valuable tool for advancing conversation head generation. We expect this metric could facilitate new advances in conversational head generation. Project Page: https://https://github.com/dc3ea9f/PreferenceScore.

Interactive Conversational Head Generation

We introduce a new conversation head generation benchmark for synthesizing behaviors of a single interlocutor in a face-to-face conversation. The capability to automatically synthesize interlocutors which can participate in long and multi-turn conversations is vital and offer benefits for various applications, including digital humans, virtual agents, and social robots. While existing research primarily focuses on talking head generation (one-way interaction), hindering the ability to create a digital human for conversation (two-way) interaction due to the absence of listening and interaction parts. In this work, we construct two datasets to address this issue, ``ViCo'' for independent talking and listening head generation tasks at the sentence level, and ``ViCo-X'', for synthesizing interlocutors in multi-turn conversational scenarios. Based on ViCo and ViCo-X, we define three novel tasks targeting the interaction modeling during the face-to-face conversation: 1) responsive listening head generation making listeners respond actively to the speaker with non-verbal signals, 2) expressive talking head generation guiding speakers to be aware of listeners' behaviors, and 3) conversational head generation to integrate the talking/listening ability in one interlocutor. Along with the datasets, we also propose corresponding baseline solutions to the three aforementioned tasks. Experimental results show that our baseline method could generate responsive and vivid agents that can collaborate with real person to fulfil the whole conversation. Project page: https://vico.solutions/.

Deep Equilibrium Multimodal Fusion

Multimodal fusion integrates the complementary information present in multiple modalities and has gained much attention recently. Most existing fusion approaches either learn a fixed fusion strategy during training and inference, or are only capable of fusing the information to a certain extent. Such solutions may fail to fully capture the dynamics of interactions across modalities especially when there are complex intra- and inter-modality correlations to be considered for informative multimodal fusion. In this paper, we propose a novel deep equilibrium (DEQ) method towards multimodal fusion via seeking a fixed point of the dynamic multimodal fusion process and modeling the feature correlations in an adaptive and recursive manner. This new way encodes the rich information within and across modalities thoroughly from low level to high level for efficacious downstream multimodal learning and is readily pluggable to various multimodal frameworks. Extensive experiments on BRCA, MM-IMDB, CMU-MOSI, SUN RGB-D, and VQA-v2 demonstrate the superiority of our DEQ fusion. More remarkably, DEQ fusion consistently achieves state-of-the-art performance on multiple multimodal benchmarks. The code will be released.

Visual-Aware Text-to-Speech

Dynamically synthesizing talking speech that actively responds to a listening head is critical during the face-to-face interaction. For example, the speaker could take advantage of the listener's facial expression to adjust the tones, stressed syllables, or pauses. In this work, we present a new visual-aware text-to-speech (VA-TTS) task to synthesize speech conditioned on both textual inputs and sequential visual feedback (e.g., nod, smile) of the listener in face-to-face communication. Different from traditional text-to-speech, VA-TTS highlights the impact of visual modality. On this newly-minted task, we devise a baseline model to fuse phoneme linguistic information and listener visual signals for speech synthesis. Extensive experiments on multimodal conversation dataset ViCo-X verify our proposal for generating more natural audio with scenario-appropriate rhythm and prosody.

Learning cross space mapping via DNN using large scale click-through logs

The gap between low-level visual signals and high-level semantics has been progressively bridged by continuous development of deep neural network (DNN). With recent progress of DNN, almost all image classification tasks have achieved new records of accuracy. To extend the ability of DNN to image retrieval tasks, we proposed a unified DNN model for image-query similarity calculation by simultaneously modeling image and query in one network. The unified DNN is named the cross space mapping (CSM) model, which contains two parts, a convolutional part and a query-embedding part. The image and query are mapped to a common vector space via these two parts respectively, and image-query similarity is naturally defined as an inner product of their mappings in the space. To ensure good generalization ability of the DNN, we learn weights of the DNN from a large number of click-through logs which consists of 23 million clicked image-query pairs between 1 million images and 11.7 million queries. Both the qualitative results and quantitative results on an image retrieval evaluation task with 1000 queries demonstrate the superiority of the proposed method.

Visualizing and Understanding Patch Interactions in Vision Transformer

Vision Transformer (ViT) has become a leading tool in various computer vision tasks, owing to its unique self-attention mechanism that learns visual representations explicitly through cross-patch information interactions. Despite having good success, the literature seldom explores the explainability of vision transformer, and there is no clear picture of how the attention mechanism with respect to the correlation across comprehensive patches will impact the performance and what is the further potential. In this work, we propose a novel explainable visualization approach to analyze and interpret the crucial attention interactions among patches for vision transformer. Specifically, we first introduce a quantification indicator to measure the impact of patch interaction and verify such quantification on attention window design and indiscriminative patches removal. Then, we exploit the effective responsive field of each patch in ViT and devise a window-free transformer architecture accordingly. Extensive experiments on ImageNet demonstrate that the exquisitely designed quantitative method is shown able to facilitate ViT model learning, leading the top-1 accuracy by 4.28% at most. Moreover, the results on downstream fine-grained recognition tasks further validate the generalization of our proposal.

Freeform Body Motion Generation from Speech

People naturally conduct spontaneous body motions to enhance their speeches while giving talks. Body motion generation from speech is inherently difficult due to the non-deterministic mapping from speech to body motions. Most existing works map speech to motion in a deterministic way by conditioning on certain styles, leading to sub-optimal results. Motivated by studies in linguistics, we decompose the co-speech motion into two complementary parts: pose modes and rhythmic dynamics. Accordingly, we introduce a novel freeform motion generation model (FreeMo) by equipping a two-stream architecture, i.e., a pose mode branch for primary posture generation, and a rhythmic motion branch for rhythmic dynamics synthesis. On one hand, diverse pose modes are generated by conditional sampling in a latent space, guided by speech semantics. On the other hand, rhythmic dynamics are synced with the speech prosody. Extensive experiments demonstrate the superior performance against several baselines, in terms of motion diversity, quality and syncing with speech. Code and pre-trained models will be publicly available through https://github.com/TheTempAccount/Co-Speech-Motion-Generation.