Libra: Building Decoupled Vision System on Large Language Models

In this work, we introduce Libra, a prototype model with a decoupled vision system on a large language model (LLM). The decoupled vision system decouples inner-modal modeling and cross-modal interaction, yielding unique visual information modeling and effective cross-modal comprehension. Libra is trained through discrete auto-regressive modeling on both vision and language inputs. Specifically, we incorporate a routed visual expert with a cross-modal bridge module into a pretrained LLM to route the vision and language flows during attention computing to enable different attention patterns in inner-modal modeling and cross-modal interaction scenarios. Experimental results demonstrate that the dedicated design of Libra achieves a strong MLLM baseline that rivals existing works in the image-to-text scenario with merely 50 million training data, providing a new perspective for future multimodal foundation models. Code is available at https://github.com/YifanXu74/Libra.

HiVG: Hierarchical Multimodal Fine-grained Modulation for Visual Grounding

Visual grounding, which aims to ground a visual region via natural language, is a task that heavily relies on cross-modal alignment. Existing works utilized uni-modal pre-trained models to transfer visual/linguistic knowledge separately while ignoring the multimodal corresponding information. Motivated by recent advancements in contrastive language-image pre-training and low-rank adaptation (LoRA) methods, we aim to solve the grounding task based on multimodal pre-training. However, there exists significant task gaps between pre-training and grounding. Therefore, to address these gaps, we propose a concise and efficient hierarchical multimodal fine-grained modulation framework, namely HiVG. Specifically, HiVG consists of a multi-layer adaptive cross-modal bridge and a hierarchical multimodal low-rank adaptation (Hi LoRA) paradigm. The cross-modal bridge can address the inconsistency between visual features and those required for grounding, and establish a connection between multi-level visual and text features. Hi LoRA prevents the accumulation of perceptual errors by adapting the cross-modal features from shallow to deep layers in a hierarchical manner. Experimental results on five datasets demonstrate the effectiveness of our approach and showcase the significant grounding capabilities as well as promising energy efficiency advantages. The project page: https://github.com/linhuixiao/HiVG.

StoryImager: A Unified and Efficient Framework for Coherent Story Visualization and Completion

Story visualization aims to generate a series of realistic and coherent images based on a storyline. Current models adopt a frame-by-frame architecture by transforming the pre-trained text-to-image model into an auto-regressive manner. Although these models have shown notable progress, there are still three flaws. 1) The unidirectional generation of auto-regressive manner restricts the usability in many scenarios. 2) The additional introduced story history encoders bring an extremely high computational cost. 3) The story visualization and continuation models are trained and inferred independently, which is not user-friendly. To these ends, we propose a bidirectional, unified, and efficient framework, namely StoryImager. The StoryImager enhances the storyboard generative ability inherited from the pre-trained text-to-image model for a bidirectional generation. Specifically, we introduce a Target Frame Masking Strategy to extend and unify different story image generation tasks. Furthermore, we propose a Frame-Story Cross Attention Module that decomposes the cross attention for local fidelity and global coherence. Moreover, we design a Contextual Feature Extractor to extract contextual information from the whole storyline. The extensive experimental results demonstrate the excellent performance of our StoryImager. The code is available at https://github.com/tobran/StoryImager.

Music Style Transfer with Time-Varying Inversion of Diffusion Models

With the development of diffusion models, text-guided image style transfer has demonstrated high-quality controllable synthesis results. However, the utilization of text for diverse music style transfer poses significant challenges, primarily due to the limited availability of matched audio-text datasets. Music, being an abstract and complex art form, exhibits variations and intricacies even within the same genre, thereby making accurate textual descriptions challenging. This paper presents a music style transfer approach that effectively captures musical attributes using minimal data. We introduce a novel time-varying textual inversion module to precisely capture mel-spectrogram features at different levels. During inference, we propose a bias-reduced stylization technique to obtain stable results. Experimental results demonstrate that our method can transfer the style of specific instruments, as well as incorporate natural sounds to compose melodies. Samples and source code are available at https://lsfhuihuiff.github.io/MusicTI/.

Dance-to-Music Generation with Encoder-based Textual Inversion of Diffusion Models

The harmonious integration of music with dance movements is pivotal in vividly conveying the artistic essence of dance. This alignment also significantly elevates the immersive quality of gaming experiences and animation productions. While there has been remarkable advancement in creating high-fidelity music from textual descriptions, current methodologies mainly concentrate on modulating overarching characteristics such as genre and emotional tone. They often overlook the nuanced management of temporal rhythm, which is indispensable in crafting music for dance, since it intricately aligns the musical beats with the dancers' movements. Recognizing this gap, we propose an encoder-based textual inversion technique for augmenting text-to-music models with visual control, facilitating personalized music generation. Specifically, we develop dual-path rhythm-genre inversion to effectively integrate the rhythm and genre of a dance motion sequence into the textual space of a text-to-music model. Contrary to the classical textual inversion method, which directly updates text embeddings to reconstruct a single target object, our approach utilizes separate rhythm and genre encoders to obtain text embeddings for two pseudo-words, adapting to the varying rhythms and genres. To achieve a more accurate evaluation, we propose improved evaluation metrics for rhythm alignment. We demonstrate that our approach outperforms state-of-the-art methods across multiple evaluation metrics. Furthermore, our method seamlessly adapts to in-the-wild data and effectively integrates with the inherent text-guided generation capability of the pre-trained model. Samples are available at \url{https://youtu.be/D7XDwtH1YwE}.

CreativeSynth: Creative Blending and Synthesis of Visual Arts based on Multimodal Diffusion

Large-scale text-to-image generative models have made impressive strides, showcasing their ability to synthesize a vast array of high-quality images. However, adapting these models for artistic image editing presents two significant challenges. Firstly, users struggle to craft textual prompts that meticulously detail visual elements of the input image. Secondly, prevalent models, when effecting modifications in specific zones, frequently disrupt the overall artistic style, complicating the attainment of cohesive and aesthetically unified artworks. To surmount these obstacles, we build the innovative unified framework CreativeSynth, which is based on a diffusion model with the ability to coordinate multimodal inputs and multitask in the field of artistic image generation. By integrating multimodal features with customized attention mechanisms, CreativeSynth facilitates the importation of real-world semantic content into the domain of art through inversion and real-time style transfer. This allows for the precise manipulation of image style and content while maintaining the integrity of the original model parameters. Rigorous qualitative and quantitative evaluations underscore that CreativeSynth excels in enhancing artistic images' fidelity and preserves their innate aesthetic essence. By bridging the gap between generative models and artistic finesse, CreativeSynth becomes a custom digital palette.

Hierarchical Prompts for Rehearsal-free Continual Learning

Continual learning endeavors to equip the model with the capability to integrate current task knowledge while mitigating the forgetting of past task knowledge. Inspired by prompt tuning, prompt-based methods maintain a frozen backbone and train with slight learnable prompts to minimize the catastrophic forgetting that arises due to updating a large number of backbone parameters. Nonetheless, these learnable prompts tend to concentrate on the discriminatory knowledge of the current task while ignoring past task knowledge, leading to that learnable prompts still suffering from catastrophic forgetting. This paper introduces a novel rehearsal-free paradigm for continual learning termed Hierarchical Prompts (H-Prompts), comprising three categories of prompts -- class prompt, task prompt, and general prompt. To effectively depict the knowledge of past classes, class prompt leverages Bayesian Distribution Alignment to model the distribution of classes in each task. To reduce the forgetting of past task knowledge, task prompt employs Cross-task Knowledge Excavation to amalgamate the knowledge encapsulated in the learned class prompts of past tasks and current task knowledge. Furthermore, general prompt utilizes Generalized Knowledge Exploration to deduce highly generalized knowledge in a self-supervised manner. Evaluations on two benchmarks substantiate the efficacy of the proposed H-Prompts, exemplified by an average accuracy of 87.8% in Split CIFAR-100 and 70.6% in Split ImageNet-R.

Hierarchical Augmentation and Distillation for Class Incremental Audio-Visual Video Recognition

Audio-visual video recognition (AVVR) aims to integrate audio and visual clues to categorize videos accurately. While existing methods train AVVR models using provided datasets and achieve satisfactory results, they struggle to retain historical class knowledge when confronted with new classes in real-world situations. Currently, there are no dedicated methods for addressing this problem, so this paper concentrates on exploring Class Incremental Audio-Visual Video Recognition (CIAVVR). For CIAVVR, since both stored data and learned model of past classes contain historical knowledge, the core challenge is how to capture past data knowledge and past model knowledge to prevent catastrophic forgetting. We introduce Hierarchical Augmentation and Distillation (HAD), which comprises the Hierarchical Augmentation Module (HAM) and Hierarchical Distillation Module (HDM) to efficiently utilize the hierarchical structure of data and models, respectively. Specifically, HAM implements a novel augmentation strategy, segmental feature augmentation, to preserve hierarchical model knowledge. Meanwhile, HDM introduces newly designed hierarchical (video-distribution) logical distillation and hierarchical (snippet-video) correlative distillation to capture and maintain the hierarchical intra-sample knowledge of each data and the hierarchical inter-sample knowledge between data, respectively. Evaluations on four benchmarks (AVE, AVK-100, AVK-200, and AVK-400) demonstrate that the proposed HAD effectively captures hierarchical information in both data and models, resulting in better preservation of historical class knowledge and improved performance. Furthermore, we provide a theoretical analysis to support the necessity of the segmental feature augmentation strategy.

Three Heads Are Better Than One: Complementary Experts for Long-Tailed Semi-supervised Learning

We address the challenging problem of Long-Tailed Semi-Supervised Learning (LTSSL) where labeled data exhibit imbalanced class distribution and unlabeled data follow an unknown distribution. Unlike in balanced SSL, the generated pseudo-labels are skewed towards head classes, intensifying the training bias. Such a phenomenon is even amplified as more unlabeled data will be mislabeled as head classes when the class distribution of labeled and unlabeled datasets are mismatched. To solve this problem, we propose a novel method named ComPlementary Experts (CPE). Specifically, we train multiple experts to model various class distributions, each of them yielding high-quality pseudo-labels within one form of class distribution. Besides, we introduce Classwise Batch Normalization for CPE to avoid performance degradation caused by feature distribution mismatch between head and non-head classes. CPE achieves state-of-the-art performances on CIFAR-10-LT, CIFAR-100-LT, and STL-10-LT dataset benchmarks. For instance, on CIFAR-10-LT, CPE improves test accuracy by over >2.22% compared to baselines. Code is available at https://github.com/machengcheng2016/CPE-LTSSL.

Erasing Self-Supervised Learning Backdoor by Cluster Activation Masking

Researchers have recently found that Self-Supervised Learning (SSL) is vulnerable to backdoor attacks. The attacker can embed hidden SSL backdoors via a few poisoned examples in the training dataset and maliciously manipulate the behavior of downstream models. To defend against SSL backdoor attacks, a feasible route is to detect and remove the poisonous samples in the training set. However, the existing SSL backdoor defense method fails to detect the poisonous samples precisely. In this paper, we propose to erase the SSL backdoor by cluster activation masking and propose a novel PoisonCAM method. After obtaining the threat model trained on the poisoned dataset, our method can precisely detect poisonous samples based on the assumption that masking the backdoor trigger can effectively change the activation of a downstream clustering model. In experiments, our PoisonCAM achieves 96% accuracy for backdoor trigger detection compared to 3% of the state-of-the-art method on poisoned ImageNet-100. Moreover, our proposed PoisonCAM significantly improves the performance of the trained SSL model under backdoor attacks compared to the state-of-the-art method. Our code will be available at https://github.com/LivXue/PoisonCAM.