ACM MM 2024 论文列表

Recently, diffusion-based depth estimation methods have drawn widespread attention due to their elegant denoising patterns and promising performance. However, they are typically unreliable under adverse conditions prevalent in real-world scenarios, such as rainy, snowy, etc. In this paper, we propose a novel robust depth estimation method called D4RD, featuring a custom contrastive learning mode tailored for diffusion models to mitigate performance degradation in complex environments. Concretely, we integrate the strength of knowledge distillation into contrastive learning, building the `trinity' contrastive scheme. This scheme utilizes the sampled noise of the forward diffusion process as a natural reference, guiding the predicted noise in diverse scenes toward a more stable and precise optimum. Moreover, we extend noise-level trinity to encompass more generic feature and image levels, establishing a multi-level contrast to distribute the burden of robust perception across the overall network. Before addressing complex scenarios, we enhance the stability of the baseline diffusion model with three straightforward yet effective improvements, which facilitate convergence and remove depth outliers. Extensive experiments demonstrate that D4RD surpasses existing state-of-the-art solutions on synthetic corruption datasets and real-world weather conditions. Source code and data are available at \url{https://github.com/wangjiyuan9/D4RD}.

Detecting diffusion-generated images has recently developed as an emerging research area. Existing diffusion-based datasets predominantly focus on general image generation. However, facial forgeries, which pose severe social risks, have remained less explored thus far. To address this gap, this paper introduces DiFF, a comprehensive dataset dedicated to face-focused diffusion-generated images. DiFF comprises over 500,000 images that are synthesized using thirteen distinct generation methods under four conditions. In particular, this dataset utilizes 30,000 carefully collected textual and visual prompts, ensuring the synthesis of images with both high fidelity and semantic consistency. We conduct extensive experiments on the DiFF dataset via human subject tests and several representative forgery detection methods. The results demonstrate that the binary detection accuracies of both human observers and automated detectors often fall below 30%, revealing insights on the challenges in detecting diffusion-generated facial forgeries. Moreover, our experiments demonstrate that DiFF, compared to previous facial forgery datasets, contains a more diverse and realistic range of forgeries, showcasing its potential to aid in the development of more generalized detectors. Finally, we propose an edge graph regularization approach to effectively enhance the generalization capability of existing detectors.

Federated Learning (FL) is an emerging direction in distributed machine learning that enables jointly training a global model without sharing data with server. However, data heterogeneity biases the parameter aggregation at the server, leading to slower convergence and poorer accuracy of the global model. To cope with this, most of the existing works involve enforcing regularization in local optimization or improving the model aggregation scheme at the server. Though effective, they lack a deep understanding of cross-client features. In this paper, we propose a saliency latent space feature aggregation method (FedSLS) across federated clients. By Guided BackPropagation (GBP), we transform deep models into powerful and flexible visual fidelity encoders, applicable to general state inputs across different image domains, and achieve powerful aggregation in the form of saliency latent features. Notably, since GBP is label-insensitive, it is sufficient to capture saliency features only once on each client. Experimental results demonstrate that FedSLS leads to significant improvements over the state-of-the-arts in terms of accuracies, especially in highly heterogeneous settings. For example, on CIFAR-10 dataset, FedSLS achieves 63.43% accuracy within the strongly heterogeneous environment α=0.05, which is 6% to 23% higher than other baselines.

LiDAR-based 3D object detection has seen impressive advances in recent times. However, deploying trained 3D detectors in the real world often yields unsatisfactory performance when the distribution of the test data significantly deviates from the training data due to different weather conditions, object sizes, etc. A key factor in this performance degradation is the diminished generalizability of pre-trained models, which creates a sharp loss landscape during training. Such sharpness, when encountered during testing, can precipitate significant performance declines, even with minor data variations. To address the aforementioned challenges, we propose dual-perturbation optimization (DPO) for Test-time Adaptation in 3D Object Detection (TTA-3OD). We minimize the sharpness to cultivate a flat loss landscape to ensure model resiliency to minor data variations, thereby enhancing the generalization of the adaptation process. To fully capture the inherent variability of the test point clouds, we further introduce adversarial perturbation to the input BEV features to better simulate the noisy test environment. As the dual perturbation strategy relies on trustworthy supervision signals, we utilize a reliable Hungarian matcher to filter out pseudo-labels sensitive to perturbations. Additionally, we introduce early Hungarian cutoff to avoid error accumulation from incorrect pseudo-labels by halting the adaptation process. Extensive experiments across three types of transfer tasks demonstrate that the proposed DPO significantly surpasses previous state-of-the-art approaches, specifically on Waymo → KITTI, outperforming the most competitive baseline by 57.72% in AP3D and reaching 91% of the fully supervised upper bound. Our code is available at https://github.com/Jo-wang/DPO.

Video hashing is a technique of encoding videos into binary vectors, facilitating efficient video storage and high-speed computation. Current approaches to video hashing predominantly utilize sequential frame images to produce semantic binary codes. However, videos encompass not only visual but also audio signals. Therefore, we propose a tri-level Transformer-based audio-visual hashing technique for video retrieval, named AVHash. It first processes audio and visual signals separately using pre-trained AST and ViT large models, and then projects temporal audio and keyframes into a shared latent semantic space using a Transformer encoder. Subsequently, a gated attention mechanism is designed to fuse the paired audio-visual signals in the video, followed by another Transformer encoder leading to the final video representation. The training of this AVHash model is directed by a video-based contrastive loss as well as a semantic alignment regularization term for audio-visual signals. Experimental results show that AVHash significantly outperforms existing video hashing methods in video retrieval tasks. Furthermore, ablation studies reveal that while video hashing based solely on visual signals achieves commendable mAP scores, the incorporation of audio signals can further boost its performance for video retrieval.

Change captioning involves describing the subtle changes between a pair of similar images. Although existing efforts have achieved compelling success, they overlook the potential of multimodal large language models (MLLMs) in tackling this challenging task. In this work, we aim to empower MLLMs with the capability to perceive subtle differences between paired images and enhance their performance in generating change captions. Specifically, we present a diFferentIal-perceptive aNd rEtRieval-augmented MLLM (FINER-MLLM) tailored for this task. In particular, FINER-MLLM leverages LoRA fine-tuned MLLM's image encoder to extract image patch features, enabling the capture of detailed image information. Subsequently, within MLLM's feature extraction, typically Q-Former, FINER-MLLM incorporates dual constraints: the intra-image feature independence constraint and the inter-image feature alignment constraint. These constraints ensure that the features can comprehensively extract subtle visual information within each image and that corresponding features across images align effectively. Last, we introduced the retrieval augmentation to first retrieve the relevant corpus to facilitate the MLLM's decoder i.e., LLM, in generating accurate change captions. Extensive experiments on three benchmark datasets, i.e., CLEVR-Change, Spot-the-Diff, and Image-Editing-Request, demonstrate the superiority of our proposed method.

Building on recent breakthroughs in diffusion-based text-to-image synthesis (TIS), training-free text-guided image editing (TIE) has emerged as an indispensable aspect of modern image editing practices. This technique involves the modification of features within attention layers to alter objects or their attributes within images during the generation process. Despite its utility, current image editing algorithms face challenges, particularly when editing multiple objects in an image. In this paper, we introduce VICTORIA, a novel approach that augments TIE by incorporating linguistic knowledge into the manipulation of attention maps during image generation. VICTORIA capitalizes on mechanisms within self-attention layers to ensure spatial consistency between source and target images. Further, we design a novel loss function that refines cross-attention maps, ensuring their alignment with linguistic constraints, thereby enhancing the editing precision of multiple target objects. We also present a linguistic mask blending technique that aids in the retention of information in regions not subjected to modification. Experimental results across seven diverse datasets show that VICTORIA achieves significant improvements over state-of-the-art methods. Our work underscores the critical role and effectiveness of linguistic analysis in elevating the performance of TIE, with a specific emphasis on multi-object scenarios. The code is available at https://github.com/alibaba/EasyNLP/tree/master/diffusion/VICTORIA.

Fine-grained image retrieval (FGIR) is to learn visual representations that distinguish visually similar objects while maintaining generalization. Existing methods propose to generate discriminative features, but rarely consider the particularity of the FGIR task itself. This paper presents a meticulous analysis leading to the proposal of practical guidelines to identify subcategory-specific discrepancies and generate discriminative features to design effective FGIR models. These guidelines include emphasizing the object (G1), highlighting subcategory-specific discrepancies (G2), and employing effective training strategy (G3). Following G1 and G2, we design a novel Dual Visual Filtering mechanism for the plain visual transformer, denoted as DVF, to capture subcategory-specific discrepancies. Specifically, the dual visual filtering mechanism comprises an object-oriented module and a semantic-oriented module. These components serve to magnify objects and identify discriminative regions, respectively. Following G3, we implement a discriminative model training strategy to improve the discriminability and generalization ability of DVF. Extensive analysis and ablation studies confirm the efficacy of our proposed guidelines. Without bells and whistles, the proposed DVF achieves state-of-the-art performance on three widely-used fine-grained datasets in closed-set and open-set settings.

Although multi-view learning has achieved remarkable progress over the past decades, most existing methods implicitly assume that all views (or modalities) are well-aligned. In practice, however, collecting fully aligned views is challenging due to complexities and discordances in time and space, resulting in the Partially View-unaligned Problem (PVP), such as audio-video asynchrony caused by network congestion. While some methods are proposed to align the unaligned views by learning view-invariant representations, almost all of them overlook specific information across different views for complementarity, limiting performance improvement. To address these problems, we propose a robust framework, dubbed VariatIonal ConTrAstive Learning (VITAL), designed to learn both common and specific information simultaneously. To be specific, each data sample is first modeled as a Gaussian distribution in the latent space, where the mean estimates the most probable common information, while the variance indicates view-specific information. Second, by using variational inference, VITAL conducts intra- and inter-view contrastive learning to preserve common and specific semantics in the distribution representations, thereby achieving comprehensive perception. As a result, the common representation (mean) could be used to guide category-level realignment, while the specific representation (variance) complements sample semantic information, thereby boosting overall performance. Finally, considering the abundance of False Negative Pairs (FNPs) generated by unsupervised contrastive learning, we propose a robust loss function that seamlessly incorporates FNP rectification into the contrastive learning paradigm. Empirical evaluations on eight benchmark datasets reveal that VITAL outperforms ten state-of-the-art deep clustering baselines, demonstrating its efficacy in both partially and fully aligned scenarios. The Code is available at https://github.com/He-Changhao/2024-MM-VITAL.

Shadow, as a natural consequence of light interacting with objects, plays a crucial role in shaping the aesthetics of an image, which however also impairs the content visibility and overall visual quality. Recent shadow removal approaches employ the mechanism of attention, due to its effectiveness, as a key component. However, they often suffer from two issues including large model size and high computational complexity for practical use. To address these shortcomings, this work devises a lightweight yet accurate shadow removal framework. First, we analyze the characteristics of the shadow removal task to seek the key information required for reconstructing shadow regions and designing a novel regional attention mechanism to effectively capture such information. Then, we customize a Regional Attention Shadow Removal Model (RASM, in short), which leverages non-shadow areas to assist in restoring shadow ones. Unlike existing attention-based models, our regional attention strategy allows each shadow region to interact more rationally with its surrounding non-shadow areas, for seeking the regional contextual correlation between shadow and non-shadow areas. Extensive experiments are conducted to demonstrate that our proposed method delivers superior performance over other state-of-the-art models in terms of accuracy and efficiency, making it appealing for practical applications. Our code can be found at https://github.com/CalcuLuUus/RASM.

Visual grounding is a task of locating the object referred by a natural language description. To reduce annotation costs, recent researchers are devoted into one-stage weakly supervised methods for visual grounding, which typically adopt the anchor-text matching paradigm. Despite the efficiency, we identify that anchor representations are often noisy and insufficient to describe object information, which inevitably hinders the vision-language alignments. In this paper, we propose a novel query-based one-stage framework for weakly supervised visual grounding, namely QueryMatch. Different from previous work, QueryMatch represents candidate objects with a set of query features, which inherently establish accurate one-to-one associations with visual objects. In this case, QueryMatch re-formulates weakly supervised visual grounding as a query-text matching problem, which can be optimized via the query-based contrastive learning. Based on QueryMatch, we further propose an innovative strategy for effective weakly supervised learning, namely Active Query Selection (AQS). In particular, AQS aims to enhance the effectiveness of query-based contrastive learning by actively selecting high-quality query features. Through this strategy, AQS can greatly benefit the weakly supervised learning of QueryMatch. To validate our approach, we conduct extensive experiments on three benchmark datasets of two grounding tasks, i.e., referring expression comprehension (REC) and segmentation (RES). Experimental results not only show the state-of-art performance of QueryMatch in two tasks, e.g., over +5% [email protected] on RefCOCO in REC and over +20% mIOU on RefCOCO in RES, but also confirm the effectiveness of AQS in weakly supervised learning. Source codes are available at https://github.com/TensorThinker/QueryMatch.

Conversational Speech Synthesis (CSS) aims to express a target utterance with the proper speaking style in a user-agent conversation setting. Existing CSS methods employ effective multi-modal context modeling techniques to achieve empathy understanding and expression. However, they often need to design complex network architectures and meticulously optimize the modules within them. In addition, due to the limitations of small-scale datasets containing scripted recording styles, they often fail to simulate real natural conversational styles. To address the above issues, we propose a novel generative expressive CSS system, termed GPT-Talker.We transform the multimodal information of the multi-turn dialogue history into discrete token sequences and seamlessly integrate them to form a comprehensive user-agent dialogue context. Leveraging the power of GPT, we predict the token sequence, that includes both semantic and style knowledge, of response for the agent. After that, the expressive conversational speech is synthesized by the conversation-enriched VITS to deliver feedback to the user.Furthermore, we propose a large-scale Natural CSS Dataset called NCSSD, that includes both naturally recorded conversational speech in improvised styles and dialogues extracted from TV shows. It encompasses both Chinese and English languages, with a total duration of 236 hours. We conducted comprehensive experiments on the reliability of the NCSSD and the effectiveness of our GPT-Talker. Both subjective and objective evaluations demonstrate that our model outperforms other state-of-the-art CSS systems significantly in terms of naturalness and expressiveness. The Code, Dataset, and Pre-trained Model are available at: https://github.com/AI-S2-Lab/GPT-Talker.

Text recognition is an inherent integration of vision and language, encompassing the visual texture in stroke patterns and the semantic context among the character sequences. Towards advanced text recognition, there are three key challenges: (1) an encoder capable of representing the visual and semantic distributions; (2) a decoder that ensures the alignment between vision and semantics; and (3) consistency in the framework during pre-training, if it exists, and fine-tuning. Inspired by masked autoencoding, a successful pre-training strategy in both vision and language, we propose an innovative scene text recognition approach, named VL-Reader. The novelty of the VL-Reader lies in the pervasive interplay between vision and language throughout the entire process. Concretely, we first introduce a Masked Visual-Linguistic Reconstruction (MVLR) objective, which aims at simultaneously modeling visual and linguistic information. Then, we design a Masked Visual-Linguistic Decoder (MVLD) to further leverage masked vision-language context and achieve bi-modal feature interaction. The architecture of VL-Reader maintains consistency from pre-training to fine-tuning. In the pre-training stage, VL-Reader reconstructs both masked visual and text tokens, while in the fine-tuning stage, the network degrades to reconstruct all characters from an image without any masked regions. VL-reader achieves an average accuracy of 97.1% on six typical datasets, surpassing the SOTA by 1.1%. The improvement was even more significant on challenging datasets. The results demonstrate that vision and language reconstructor can serve as an effective scene text recognizer.

With the recent burst of 2D and 3D data, cross-modal retrieval has attracted increasing attention recently. However, manual labeling by non-experts will inevitably introduce corrupted annotations given ambiguous 2D/3D content. Though previous works have addressed this issue by designing a naive division strategy with hand-crafted thresholds, their performance generally exhibits great sensitivity to the threshold value. Besides, they fail to fully utilize the valuable supervisory signals within each divided subset. To tackle this problem, we propose a Divide-and-conquer 2D-3D cross-modal Alignment and Correction framework (DAC), which comprises Multimodal Dynamic Division (MDD) and Adaptive Alignment and Correction (AAC). Specifically, the former performs accurate sample division by adaptive credibility modeling for each sample based on the compensation information within multimodal loss distribution. Then in AAC, samples in distinct subsets are exploited with different alignment strategies to fully enhance the semantic compactness and meanwhile alleviate over-fitting to noisy labels, where a self-correction strategy is introduced to improve the quality of representation. Moreover. To evaluate the effectiveness in real-world scenarios, we introduce a challenging noisy benchmark, namely Objaverse-N200, which comprises 200k-level samples annotated with 1156 realistic noisy labels. Extensive experiments on both traditional and the newly proposed benchmarks demonstrate the generality and superiority of our DAC, where DAC outperforms state-of-the-art models by a large margin. (i.e., with +5.9% gain on ModelNet40 and +5.8% on Objaverse-N200).

Incorporating domain-specific visual information into text poses one of the critical challenges for domain-specific multi-modal neural machine translation (DMNMT). While most existing DMNMT methods often borrow multi-modal fusion frameworks from multi-modal neural machine translation (MNMT) in the general domain, they overlook the domain gaps between general and specific domains. Visual-to-textual interaction in a specific domain frequently exhibits multi-focus characteristics, making it difficult to consistently focus on domain-specific multi-visual details using traditional multi-modal fusion frameworks. This challenge can lead to a decrease in machine translation performance for domain-specific terms. To tackle this problem, this paper presents a virtual visual scene-guided domain-shadow multi-modal fusion mechanism to simultaneously integrate multi-grained domain visual details and text with the guidance of modality-agnostic virtual visual scene, thereby enhancing machine translation performance for DMNMT, especially for domain terms. Specifically, we first adopt a modality-mixing selection-voting strategy to generate modality-mixed domain representations through layer-by-layer intra-modality selection and inter-modality exchanging. Then, we gradually aggregate modality-mixed domain representations and text across modality boundaries with the guidance of a modality-agnostic virtual visual scene to enhance the collaboration between domain characteristics and textual semantics. The experimental results on three benchmark datasets demonstrate that our proposed approach outperforms the state-of-the-art (SOTA) methods in all machine translation tasks. The in-depth analysis further highlights the robustness and generalizability of our approach across various scenarios. Our code is available on https://github.com/HZY2023/VVDF.

Facial Expression Recognition (FER) holds significant importance in human-computer interactions. Existing cross-domain FER methods often transfer knowledge solely from a single labeled source domain to an unlabeled target domain, neglecting the comprehensive information across multiple sources. Nevertheless, cross-multidomain FER (CMFER) is very challenging for (i) the inherent inter-domain shifts across multiple domains and (ii) the intra-domain shifts stemming from the ambiguous expressions and low inter-class distinctions. In this paper, we propose a novel Learning with Alignments CMFER framework, named LA-CMFER, to handle both inter- and intra-domain shifts. Specifically, LA-CMFER is constructed with a global branch and a local branch to extract features from the full images and local subtle expressions, respectively. Based on this, LA-CMFER presents a dual-level inter-domain alignment method to force the model to prioritize hard-to-align samples in knowledge transfer at a sample level while gradually generating a well-clustered feature space with the guidance of class attributes at a cluster level, thus narrowing the inter-domain shifts. To address the intra-domain shifts, LA-CMFER introduces a multi-view intra-domain alignment method with a multi-view clustering consistency constraint where a prediction similarity matrix is built to pursue consistency between the global and local views, thus refining pseudo labels and eliminating latent noise. Extensive experiments on six benchmark datasets have validated the superiority of our LA-CMFER.