ACM MM 2024 Papers

Emotion recognition based on electroencephalogram (EEG) has garnered increasing attention in recent years due to non-invasiveness and high reliability of EEG measurements. Despite the promising performance achieved by numerous existing methods, several challenges persist. Firstly, there is the challenge of emotional label noise, stemming from the assumption that emotions remain consistently evoked and stable throughout the entirety of video observation. Such an assumption proves difficult to uphold in practical experimental settings, leading to discrepancies between EEG signals and anticipated emotional states. In addition, there's a need for comprehensive capture of temporal-spatial-spectral characteristics of EEG signals and cope with low signal-to-noise ratio (SNR) issues. To tackle these challenges, we propose a comprehensive pipeline named REmoNet, which leverages novel self-supervised techniques and multi-regularized co-learning. Two self-supervised methods, including masked channel modeling via temporal-spectral transformation and emotion contrastive learning, are introduced to facilitate the comprehensive understanding and extraction of emotion-relevant EEG representations during pre-training. Additionally, fine-tuning with multi-regularized co-learning exploits feature-dependent information through intrinsic similarity, resulting in mitigating emotional label noise. Experimental evaluations on two public datasets demonstrate that our proposed approach, REmoNet, surpasses existing state-of-the-art methods, showcasing its effectiveness in simultaneously addressing raw EEG signals and noisy emotional labels.

Skeleton-based zero-shot action recognition aims to recognize unknown human actions based on the learned priors of the known skeleton-based actions and a semantic descriptor space shared by both known and unknown categories. However, previous works mainly focus on establishing the bridges between the known skeleton representation space and semantic descriptions space at the coarse-grained level for recognizing unknown action categories, ignoring the fine-grained alignment of these two spaces, resulting in suboptimal performance in distinguishing high-similarity action categories. To address these challenges, we propose a novel method via Side information and dual-prompTs learning for skeleton-based zero-shot Action Recognition (STAR) at the fine-grained level. Specifically, 1) we decompose the skeleton into several parts based on its topology structure and introduce the side information concerning multi-part descriptions of human body movements for alignment between the skeleton and the semantic space at the fine-grained level; 2) we design the visual-attribute and semantic-part prompts to improve the intra-class compactness within the skeleton space and inter-class separability within the semantic space, respectively, to distinguish the high-similarity actions. Extensive experiments show that our method achieves state-of-the-art performance in ZSL and GZSL settings on NTU RGB+D, NTU RGB+D 120, and PKU-MMD datasets.

Deep network-based image Compressive Sensing (CS) has attracted much attention in recent years. However, there still exist the following two issues: 1) Existing methods typically use fixed-scale sampling, which leads to limited insights into the image content. 2) Most pre-trained models can only handle fixed sampling rates and fixed block scales, which restricts the scalability of the model. In this paper, we propose a novel scale-aware scalable CS network (dubbed S2-CSNet), which achieves scale-aware adaptive sampling, fine granular scalability and high-quality reconstruction with one single model. Specifically, to enhance the scalability of the model, a structural sampling matrix with a predefined order is first designed, which is a universal sampling matrix that can sample multi-scale image blocks with arbitrary sampling rates. Then, based on the universal sampling matrix, a distortion-guided scale-aware scheme is presented to achieve scale-variable adaptive sampling, which predicts the reconstruction distortion at different sampling scales from the measurements and select the optimal division scale for sampling. Furthermore, a multi-scale hierarchical sub-network under a well-defined compact framework is put forward to reconstruct the image. In the multi-scale feature domain of the sub-network, a dual spatial attention is developed to explore the local and global affinities between dense feature representations for deep fusion. Extensive experiments manifest that the proposed S2-CSNet outperforms existing state-of-the-art CS methods.

3D occupancy prediction (OCC) aims to estimate and predict the semantic occupancy state of the surrounding environment, which is crucial for scene understanding and reconstruction in the real world. However, existing methods for 3D OCC mainly rely on surround-view camera images, whose performance is still insufficient in some challenging scenarios, such as low-light conditions. To this end, we propose a new multi-modal fusion network for 3D occupancy prediction by fusing features of LiDAR point clouds and surround-view images, called FusionOcc. Our model fuses features of these two modals in 2D and 3D space, respectively. By integrating the depth information from point clouds, a cross-modal fusion module is designed to predict a 2D dense depth map, enabling an accurate depth estimation and a better transition of 2D image features into 3D space. In addition, features of voxelized point clouds are aligned and merged with image features converted by a view-transformer in 3D space. Experiments show that FusionOcc establishes the new state of the art on Occ3D-nuScenes dataset, achieving a mIoU score of 35.94% (without visibility mask) and 56.62% (with visibility mask), showing an average improvement of 3.42% compared to the best previous method. Our work provides a new baseline for further research in multi-modal fusion for 3D occupancy prediction. Codes will be made publicly at https://github.com/ShuoZhang-code/FusionOcc.

With the rise of "Metaverse" and "Web 3.0", Non-Fungible Token (NFT) has emerged as a kind of pivotal digital asset, garnering significant attention. By the end of March 2024, more than 1.7 billion NFTs have been minted across various blockchain platforms. To effectively locate a desired NFT, conducting searches within a vast array of NFTs is essential. The challenge in NFT retrieval is heightened due to the high degree of similarity among different NFTs, regarding regional and semantic aspects. In this paper, we will introduce a benchmark dataset named "NFT Top1000 Visual-Text Dataset" (NFT1000), containing 7.56 million image-text pairs, and being collected from 1000 most famous PFP1 NFT collections2 by sales volume on the Ethereum blockchain. Based on this dataset and leveraging the CLIP series of pre-trained models as our foundation, we propose the dynamic masking fine-tuning scheme. This innovative approach results in a 7.4\% improvement in the top1 accuracy rate, while utilizing merely 13\% of the total training data (0.79 million vs. 6.1 million). We also propose a robust metric Comprehensive Variance Index (CVI) to assess the similarity and retrieval difficulty of visual-text pairs data. The dataset will be released as an open-source resource. For more details, please refer to: https://github.com/ShuxunoO/NFT-Net.git.

A plethora of text-guided image editing methods has recently been developed by leveraging the impressive capabilities of large-scale diffusion-based generative models especially Stable Diffusion. Despite the success of diffusion models in producing high-quality images, their application to small object generation has been limited due to difficulties in aligning cross-modal attention maps between text and these objects. Our approach offers a training-free method that significantly mitigates this alignment issue with local and global attention guidance, enhancing the model's ability to accurately render small objects in accordance with textual descriptions. We detail the methodology in our approach, emphasizing its divergence from traditional generation techniques and highlighting its advantages. What's more important is that we also provide SOEBench (Small Object Editing), a standardized benchmark for quantitatively evaluating text-based small object generation collected from MSCOCO[22] and OpenImage[18]. Preliminary results demonstrate the effectiveness of our method, showing marked improvements in the fidelity and accuracy of small object generation compared to existing models. This advancement not only contributes to the field of AI and computer vision but also opens up new possibilities for applications in various industries where precise image generation is critical.We will release our dataset on our project page: https://soebench.github.io/

Recent advancements in the field of Diffusion Transformers have substantially improved the generation of high-quality 2D images, 3D videos, and 3D shapes. However, the effectiveness of the Transformer architecture in the domain of co-speech gesture generation remains relatively unexplored, as prior methodologies have predominantly employed the Convolutional Neural Network (CNNs) or simple a few transformer layers. In an attempt to bridge this research gap, we introduce a novel Masked Diffusion Transformer for co-speech gesture generation, referred to as MDT-A2G, which directly implements the denoising process on gesture sequences. To enhance the contextual reasoning capability of temporally aligned speech-driven gestures, we incorporate a novel Masked Diffusion Transformer. This model employs a mask modeling scheme specifically designed to strengthen temporal relation learning among sequence gestures, thereby expediting the learning process and leading to coherent and realistic motions. Apart from audio, Our MDT-A2G model also integrates multi-modal information, encompassing text, emotion, and identity. Furthermore, we propose an efficient inference strategy that diminishes the denoising computation by leveraging previously calculated results, thereby achieving a speedup with negligible performance degradation. Experimental results demonstrate that MDT-A2G excels in gesture generation, boasting a learning speed that is over 6× faster than traditional diffusion transformers and an inference speed that is 5.7× than the standard diffusion model. Our code is available at https://xiaofenmao.github.io/web-project/MDT-A2G/

Scene text retrieval aims to find all images containing the query text from an image gallery. Current efforts tend to adopt an Optical Character Recognition (OCR) pipeline, which requires complicated text detection and/or recognition processes, resulting in inefficient and inflexible retrieval. Different from them, in this work we propose to explore the intrinsic potential of Contrastive Language-Image Pre-training (CLIP) for OCR-free scene text retrieval. Through empirical analysis, we observe that the main challenges of CLIP as a text retriever are: 1) limited text perceptual scale, and 2) entangled visual-semantic concepts. To this end, a novel model termed FDP (Focus, Distinguish, and Prompt) is developed. FDP first focuses on scene text via shifting the attention to the text area and probing the hidden text knowledge, and then divides the query text into content word and function word for processing, in which a semantic-aware prompting scheme and a distracted queries assistance module are utilized. Extensive experiments show that FDP significantly enhances the inference speed while achieving better or competitive retrieval accuracy compared to existing methods. Notably, on the IIIT-STR benchmark, FDP surpasses the state-of-the-art model by 4.37% with a 4 times faster speed. Furthermore, additional experiments under phrase-level and attribute-aware scene text retrieval settings validate FDP's particular advantages in handling diverse forms of query text. The source code will be available at https://github.com/Gyann-z/FDP.

Federated learning is a promising distributed training paradigm that effectively safeguards data privacy. However, it may involve significant communication costs, which hinders training efficiency. In this paper, we aim to enhance communication efficiency from a new perspective. Specifically, we request the distributed clients to find optimal model updates relative to global model parameters within predefined random noise. For this purpose, we propose Federated Masked Random Noise (FedMRN), a novel framework that enables clients to learn a 1-bit mask for each model parameter and apply masked random noise (i.e., the Hadamard product of random noise and masks) to represent model updates. To make FedMRN feasible, we propose an advanced mask training strategy, called progressive stochastic masking (PSM). After local training, each client only need to transmit local masks and a random seed to the server. Additionally, we provide theoretical guarantees for the convergence of FedMRN under both strongly convex and non-convex assumptions. Extensive experiments are conducted on four popular datasets. The results show that FedMRN exhibits superior convergence speed and test accuracy compared to relevant baselines, while attaining a similar level of accuracy as FedAvg.

Vision Transformers (ViTs) excel in extracting global information from image patches. However, their inherent limitation lies in effectively extracting information within local regions, hindering their applicability and performance. Particularly, fully supervised pre-trained ViTs, such as Vanilla ViT and CLIP, face the challenge of locality vanishing when adapting to downstream tasks. To address this, we introduce a novel LOcality-aware pRompt lEarning (LORE) method, aiming to improve the adaptation of pre-trained ViTs to downstream tasks. LORE integrates a data-driven Black Box module (i.e., a pre-trained ViT encoder) with a knowledge-driven White Box module. The White Box module is a locality-aware prompt learning mechanism to compensate for ViTs' deficiency in incorporating local information. More specifically, it begins with the design of a Locality Interaction Network (LIN), which treats an image as a neighbor graph and employs graph convolution operations to enhance local relationships among image patches. Subsequently, a Knowledge-Locality Attention (KLA) mechanism is proposed to capture critical local regions from images, learning Knowledge-Locality (K-L) prototypes utilizing relevant semantic knowledge. Afterwards, K-L prototypes guide the training of a Prompt Generator (PG) to generate locality-aware prompts for images. The locality-aware prompts, aggregating crucial local information, serve as additional input for our Black Box module. Combining pre-trained ViTs with our locality-aware prompt learning mechanism, our Black-White Box model enables the capture of both global and local information, facilitating effective downstream task adaptation. Experimental evaluations across four downstream tasks demonstrate the effectiveness and superiority of our LORE.

With the increasing prevalence of virtual assistants, multimodal conversational recommendation systems (multimodal CRS) becomes essential for boosting customer engagement, improving conversion rates, and enhancing user satisfaction. Yet conversational samples, as training data for such a system, are difficult to obtain in large quantities, particularly in new platforms. To effectively train multimodal CRS in a small data setting, we enhance data quality to make up for the small data quantity by augmenting conversations with dialogue states. We then devise an effective dialogue state encoder to bridge the semantic gap between conversation and product representations for recommendation. To further reduce the cost of dialogue state annotation, a semi-supervised learning method is developed to effectively train the dialogue state encoder with a small set of labeled conversations. In addition, we design a correlation regularisation that leverages knowledge in the multimodal product database to help align textual and visual modalities. Experiments on the dataset MMD demonstrate the effectiveness of our method. Particularly, with only 5% of the MMD training set, our method (namely SeMANTIC) obtains better NDCG scores than those of baseline models trained on the full MMD training set.

Recent generative methods have revolutionized the way of human motion synthesis, such as Variational Autoencoders (VAEs), Generative Adversarial Networks (GANs), and Denoising Diffusion Probabilistic Models (DMs). These methods have gained significant attention in human motion fields. However, there are still challenges in unconditionally generating highly diverse human motions from a given distribution. To enhance the diversity of synthesized human motions, previous methods usually employ deep neural networks (DNNs) to train a transport map that transforms Gaussian noise distribution into real human motion distribution. According to Figalli's regularity theory, the optimal transport map computed by DNNs frequently exhibits discontinuities. This is due to the inherent limitation of DNNs in representing only continuous maps. Consequently, the generated human motions tend to heavily concentrate on densely populated regions of the data distribution, resulting in mode collapse or mode mixture. To address the issues, we propose an efficient method called MOOT for unconditional human motion synthesis. First, we utilize a reconstruction network based on GRU and transformer to map human motions to latent space. Next, we employ convex optimization to match the noise distribution with the latent space distribution of human motions through the Optimal Transport (OT) map. Then, we combine the extended OT map with the generator of reconstruction network to generate new human motions. Thereby overcoming the issues of mode collapse and mode mixture. MOOT generates a latent code distribution that is well-behaved and highly structured, providing a strong motion prior for various applications in the field of human motion. Through qualitative and quantitative experiments, MOOT achieves state-of-the-art results surpassing the latest methods, validating its superiority in unconditional human motion generation.

Person re-identification (ReID) is crucial in video surveillance, aiming to match individuals across different camera views while cloth-changing person re-identification (CC-ReID) focuses on pedestrians changing attire. Many existing CC-ReID methods overlook generalization, crucial for universality across cloth-consistent and cloth-changing scenarios. This paper pioneers exploring the cloth-generalized person re-identification (CG-ReID) task and introduces the Cloth-aware Augmentation (CaAug) strategy. Comprising domain augmentation and feature augmentation, CaAug aims to learn identity-relevant features adaptable to both scenarios. Domain augmentation involves creating diverse fictitious domains and simulating various clothing scenarios. Supervising features from different cloth domains enhances robustness and generalization against clothing changes. Additionally, for feature augmentation, element exchange introduces diversity concerning clothing changes. Regularizing the model with these augmented features strengthens resilience against clothing change uncertainty. Extensive experiments on cloth-changing datasets demonstrate the efficacy of our approach, consistently outperforming state-of-the-art methods.

Prompt learning represents a promising method for adapting pre-trained vision-language models (VLMs) to various downstream tasks by learning a set of text embeddings. One challenge inherent to these methods is the poor generalization performance due to the invalidity of the learned text embeddings for unseen tasks. A straightforward approach to bridge this gap is to freeze the text embeddings in prompts, which results in a lack of capacity to adapt VLMs for downstream tasks. To address this dilemma, we propose a paradigm called EnPrompt with a novel External Layer (EnLa). Specifically, we propose a textual external layer and learnable visual embeddings for adapting VLMs to downstream tasks. The learnable external layer is built upon valid embeddings of pre-trained CLIP. This design considers the balance of learning capabilities between the two branches. To align the textual and visual features, we propose a novel two-pronged approach: i) we introduce the optimal transport as the discrepancy metric to align the vision and text modalities, and ii) we introduce a novel strengthening feature to enhance the interaction between these two modalities. Four representative experiments (i.e., base-to-novel generalization, few-shot learning, cross-dataset generalization, domain shifts generalization) across 15 datasets demonstrate that our method outperforms the existing prompt learning method.

Object detectors often suffer a decrease in performance due to the large domain gap between the training data (source domain) and real-world data (target domain). Diffusion-based generative models have shown remarkable abilities in generating high-quality and diverse images, suggesting their potential for extracting valuable feature from various domains. To effectively leverage the cross-domain feature representation of diffusion models, in this paper, we train a detector with frozen-weight diffusion model on the source domain, then employ it as a teacher model to generate pseudo labels on the unlabeled target domain, which are used to guide the supervised learning of the student model on the target domain. We refer to this approach as Diffusion Domain Teacher (DDT). By employing this straightforward yet potent framework, we significantly improve cross-domain object detection performance without compromising the inference speed. Our method achieves an average mAP improvement of 21.2% compared to the baseline on 6 datasets from three common cross-domain detection benchmarks (Cross-Camera, Syn2Real, Real2Artistic), surpassing the current state-of-the-art (SOTA) methods by an average of 5.7% mAP. Furthermore, extensive experiments demonstrate that our method consistently brings improvements even in more powerful and complex models, highlighting broadly applicable and effective domain adaptation capability of our DDT.

3D face recognition is subject to frequent spoofing attacks, in which 3D face presentation attack is one of the most notorious attacks. The attacker takes advantages of 3D scanning and printing techniques to generate masks of targets, which has found success in numerous real-life examples. The salient feature in such attacks is to obtain 3D face models through 3D scanning, though relatively more expensive and inconvenient when comparing with 2D photos. In this work, we propose a new method, DREAM, to recover 3D face models from single 2D image. Specifically, we adopt a black-box approach, which recovers 'sufficient' depths to defeat target recognition models (e.g., face identification and face authentication models) by accessing its output and the corresponding RGB photo. The key observation is that it is not necessary to restore the true value of depths, but only need to recover the essential features relevant to the target model. We used four public 3D face datasets to verify the effectiveness of DREAM. The experimental results show that DREAM can achieve a success rate of 94% on face authentication model, even in cross-dataset testing, and a success rate of 36% on face identification model.

Visual grounding is the task of locating objects specified by natural language expressions. Existing methods extend generic object detection frameworks to tackle this task. They typically extract visual and textual features separately using independent visual and textual encoders, then fuse these features in a multi-modal decoder for final prediction. However, visual grounding presents unique challenges. It often involves locating objects with different text descriptions within the same image. Existing methods struggle with this task because the independent visual encoder produces identical visual features for the same image, limiting detection performance. Some recently approaches propose various language-guided visual encoders to address this issue, but they mostly rely solely on textual information and require sophisticated designs. In this paper, we introduce Multi-modal Conditional Adaptation (MMCA), which enables the visual encoder to adaptively update weights, directing its focus towards text-relevant regions. Specifically, we first integrate information from different modalities to obtain multi-modal embeddings. Then we utilize a set of weighting coefficients, which generated from the multimodal embeddings, to reorganize the weight update matrices and apply them to the visual encoder of the visual grounding model. Extensive experiments on four widely used datasets demonstrate that MMCA achieves significant improvements and state-of-the-art results. Ablation experiments further demonstrate the lightweight and efficiency of our method. Our source code is available at: https://github.com/Mr-Bigworth/MMCA.