AAAI 2024 Papers

Weakly-supervised temporal action localization aims to locate action regions and identify action categories in untrimmed videos simultaneously by taking only video-level labels as the supervision. Pseudo label generation is a promising strategy to solve the challenging problem, but the current methods ignore the natural temporal structure of the video that can provide rich information to assist such a generation process. In this paper, we propose a novel weakly-supervised temporal action localization method by inferring salient snippet-feature. First, we design a saliency inference module that exploits the variation relationship between temporal neighbor snippets to discover salient snippet-features, which can reflect the significant dynamic change in the video. Secondly, we introduce a boundary refinement module that enhances salient snippet-features through the information interaction unit. Then, a discrimination enhancement module is introduced to enhance the discriminative nature of snippet-features. Finally, we adopt the refined snippet-features to produce high-fidelity pseudo labels, which could be used to supervise the training of the action localization network. Extensive experiments on two publicly available datasets, i.e., THUMOS14 and ActivityNet v1.3, demonstrate our proposed method achieves significant improvements compared to the state-of-the-art methods. Our source code is available at https://github.com/wuli55555/ISSF.

Mirror detection is of great significance for avoiding false recognition of reflected objects in computer vision tasks. Existing mirror detection frameworks usually follow a supervised setting, which relies heavily on high quality labels and suffers from poor generalization. To resolve this, we instead propose the first weakly-supervised mirror detection framework and also provide the first scribble-based mirror dataset. Specifically, we relabel 10,158 images, most of which have a labeled pixel ratio of less than 0.01 and take only about 8 seconds to label. Considering that the mirror regions usually show great scale variation, and also irregular and occluded, thus leading to issues of incomplete or over detection, we propose a local-global feature enhancement (LGFE) module to fully capture the context and details. Moreover, it is difficult to obtain basic mirror structure using scribble annotation, and the distinction between foreground (mirror) and background (non-mirror) features is not emphasized caused by mirror reflections. Therefore, we propose a foreground-aware mask attention (FAMA), integrating mirror edges and semantic features to complete mirror regions and suppressing the influence of backgrounds. Finally, to improve the robustness of the network, we propose a prototype contrast loss (PCL) to learn more general foreground features across images. Extensive experiments show that our network outperforms relevant state-of-the-art weakly supervised methods, and even some fully supervised methods. The dataset and codes are available at https://github.com/winter-flow/WSMD.

State-of-the-art techniques in weakly-supervised semantic segmentation (WSSS) using image-level labels exhibit severe performance degradation on driving scene datasets such as Cityscapes. To address this challenge, we develop a new WSSS framework tailored to driving scene datasets. Based on extensive analysis of dataset characteristics, we employ Contrastive Language-Image Pre-training (CLIP) as our baseline to obtain pseudo-masks. However, CLIP introduces two key challenges: (1) pseudo-masks from CLIP lack in representing small object classes, and (2) these masks contain notable noise. We propose solutions for each issue as follows. (1) We devise Global-Local View Training that seamlessly incorporates small-scale patches during model training, thereby enhancing the model's capability to handle small-sized yet critical objects in driving scenes (e.g., traffic light). (2) We introduce Consistency-Aware Region Balancing (CARB), a novel technique that discerns reliable and noisy regions through evaluating the consistency between CLIP masks and segmentation predictions. It prioritizes reliable pixels over noisy pixels via adaptive loss weighting. Notably, the proposed method achieves 51.8\% mIoU on the Cityscapes test dataset, showcasing its potential as a strong WSSS baseline on driving scene datasets. Experimental results on CamVid and WildDash2 demonstrate the effectiveness of our method across diverse datasets, even with small-scale datasets or visually challenging conditions. The code is available at https://github.com/k0u-id/CARB.

Despite weakly supervised object detection (WSOD) being a promising step toward evading strong instance-level annotations, its capability is confined to closed-set categories within a single training dataset. In this paper, we propose a novel weakly supervised open-vocabulary object detection framework, namely WSOVOD, to extend traditional WSOD to detect novel concepts and utilize diverse datasets with only image-level annotations. To achieve this, we explore three vital strategies, including dataset-level feature adaptation, image-level salient object localization, and region-level vision-language alignment. First, we perform data-aware feature extraction to produce an input-conditional coefficient, which is leveraged into dataset attribute prototypes to identify dataset bias and help achieve cross-dataset generalization. Second, a customized location-oriented weakly supervised region proposal network is proposed to utilize high-level semantic layouts from the category-agnostic segment anything model to distinguish object boundaries. Lastly, we introduce a proposal-concept synchronized multiple-instance network, i.e., object mining and refinement with visual-semantic alignment, to discover objects matched to the text embeddings of concepts. Extensive experiments on Pascal VOC and MS COCO demonstrate that the proposed WSOVOD achieves new state-of-the-art compared with previous WSOD methods in both close-set object localization and detection tasks. Meanwhile, WSOVOD enables cross-dataset and open-vocabulary learning to achieve on-par or even better performance than well-established fully-supervised open-vocabulary object detection (FSOVOD).

To enhance the interaction between intelligent systems and the environment, locating the affordance regions of objects is crucial. These regions correspond to specific areas that provide distinct functionalities. Humans often acquire the ability to identify these regions through action demonstrations and verbal instructions. In this paper, we present a novel multimodal framework that extracts affordance knowledge from exocentric images, which depict human-object interactions, as well as from accompanying textual descriptions that describe the performed actions. The extracted knowledge is then transferred to egocentric images. To achieve this goal, we propose the HOI-Transfer Module, which utilizes local perception to disentangle individual actions within exocentric images. This module effectively captures localized features and correlations between actions, leading to valuable affordance knowledge. Additionally, we introduce the Pixel-Text Fusion Module, which fuses affordance knowledge by identifying regions in egocentric images that bear resemblances to the textual features defining affordances. We employ a Weakly Supervised Multimodal Affordance (WSMA) learning approach, utilizing image-level labels for training. Through extensive experiments, we demonstrate the superiority of our proposed method in terms of evaluation metrics and visual results when compared to existing affordance grounding models. Furthermore, ablation experiments confirm the effectiveness of our approach. Code:https://github.com/xulingjing88/WSMA.

In recent years, Few-shot Object Detection (FSOD) has become an increasingly important research topic in computer vision. However, existing FSOD methods require strong annotations including category labels and bounding boxes, and their performance is heavily dependent on the quality of box annotations. However, acquiring strong annotations is both expensive and time-consuming. This inspires the study on weakly supervised FSOD (WS-FSOD in short), which realizes FSOD with only image-level annotations, i.e., category labels. In this paper, we propose a new and effective weakly supervised FSOD method named WFS-DETR. By a well-designed pretraining process, WFS-DETR first acquires general object localization and integrity judgment capabilities on large-scale pretraining data. Then, it introduces object integrity into multiple-instance learning to solve the common local optimum problem by comprehensively exploiting both semantic and visual information. Finally, with simple fine-tuning, it transfers the knowledge learned from the base classes to the novel classes, which enables accurate detection of novel objects. Benefiting from this ``pretraining-refinement'' mechanism, WSF-DETR can achieve good generalization on different datasets. Extensive experiments also show that the proposed method clearly outperforms the existing counterparts in the WS-FSOD task.

Point cloud salient object detection (PCSOD) is a newly proposed task in 3D dense segmentation. However, the acquisition of accurate 3D dense annotations comes at a high cost, severely limiting the progress of PCSOD. To address this issue, we propose the first weakly supervised PCSOD (named WeakPCSOD) model, which relies solely on cheap 3D bounding box annotations. In WeakPCSOD, we extract noise-free supervision from coarse 3D bounding boxes while mitigating shape biases inherent in box annotations. To achieve this, we introduce a novel mask-to-box (M2B) transformation and a color consistency (CC) loss. The M2B transformation, from a shape perspective, disentangles predictions from labels, enabling the extraction of noiseless supervision from labels while preserving object shapes independently of the box bias. From an appearance perspective, we further introduce the CC loss to provide dense supervision, which mitigates the non-unique predictions stemming from weak supervision and substantially reduces prediction variability. Furthermore, we employ a self-training (ST) strategy to enhance performance by utilizing high-confidence pseudo labels. Notably, the M2B transformation, CC loss, and ST strategy are seamlessly integrated into any model and incur no computational costs for inference. Extensive experiments demonstrate the effectiveness of our WeakPCSOD model, even comparable to fully supervised models utilizing dense annotations.

We propose a generalized method for boosting the generalization ability of pre-trained vision-language models (VLMs) while fine-tuning on downstream few-shot tasks. The idea is realized by exploiting out-of-distribution (OOD) detection to predict whether a sample belongs to a base distribution or a novel distribution and then using the score generated by a dedicated competition based scoring function to fuse the zero-shot and few-shot classifier. The fused classifier is dynamic, which will bias towards the zero-shot classifier if a sample is more likely from the distribution pre-trained on, leading to improved base-to-novel generalization ability. Our method is performed only in test stage, which is applicable to boost existing methods without time-consuming re-training. Extensive experiments show that even weak distribution detectors can still improve VLMs' generalization ability. Specifically, with the help of OOD detectors, the harmonic mean of CoOp and ProGrad increase by 2.6 and 1.5 percentage points over 11 recognition datasets in the base-to-novel setting.

Spatiotemporal predictive learning is a paradigm that empowers models to learn spatial and temporal patterns by predicting future frames from past frames in an unsupervised manner. This method typically uses recurrent units to capture long-term dependencies, but these units often come with high computational costs and limited performance in real-world scenes. This paper presents an innovative Wavelet-based SpatioTemporal (WaST) framework, which extracts and adaptively controls both low and high-frequency components at image and feature levels via 3D discrete wavelet transform for faster processing while maintaining high-quality predictions. We propose a Time-Frequency Aware Translator uniquely crafted to efficiently learn short- and long-range spatiotemporal information by individually modeling spatial frequency and temporal variations. Meanwhile, we design a wavelet-domain High-Frequency Focal Loss that effectively supervises high-frequency variations. Extensive experiments across various real-world scenarios, such as driving scene prediction, traffic flow prediction, human motion capture, and weather forecasting, demonstrate that our proposed WaST achieves state-of-the-art performance over various spatiotemporal prediction methods.

As attitude and motion sensing components, inertial sensors are widely used in various portable devices, covering consumer electronics, sports health, aerospace, etc. But the severe intrinsic errors of inertial sensors heavily restrain their function implementation, especially the advanced functionality, including motion trajectory recovery and motion semantic recognition, which attracts considerable attention. As a mainstream signal processing method, wavelet is hailed as the mathematical microscope of signal due to the plentiful and diverse wavelet basis functions. However, complicated noise types and application scenarios of inertial sensors make selecting wavelet basis perplexing. To this end, we propose a wavelet dynamic selection network (WDSNet), which intelligently selects the appropriate wavelet basis for variable inertial signals. In addition, existing deep learning architectures excel at extracting features from input data but neglect to learn the characteristics of target categories, which is essential to enhance the category awareness capability, thereby improving the selection of wavelet basis. Therefore, we propose a category representation mechanism (CRM), which enables the network to extract and represent category features without increasing trainable parameters. Furthermore, CRM transforms the common fully connected network into category representations, which provide closer supervision to the feature extractor than the far and trivial one-hot classification labels. We call this process of imposing interpretability on a network and using it to supervise the feature extractor the feature supervision mechanism, and its effectiveness is demonstrated experimentally and theoretically in this paper. The enhanced inertial signal can perform impracticable tasks with regard to the original signal, such as trajectory reconstruction. Both quantitative and visual results show that WDSNet outperforms the existing methods. Remarkably, WDSNet, as a weakly-supervised method, achieves the state-of-the-art performance of all the compared fully-supervised methods.

In the existing spectral GNNs, polynomial-based methods occupy the mainstream in designing a filter through the Laplacian matrix. However, polynomial combinations factored by the Laplacian matrix naturally have limitations in message passing (e.g., over-smoothing). Furthermore, most existing spectral GNNs are based on polynomial bases, which struggle to capture the high-frequency parts of the graph spectral signal. Additionally, we also find that even increasing the polynomial order does not change this situation, which means polynomial-based models have a natural deficiency when facing high-frequency signals. To tackle these problems, we propose WaveNet, which aims to effectively capture the high-frequency part of the graph spectral signal from the perspective of wavelet bases through reconstructing the message propagation matrix. We utilize Multi-Resolution Analysis (MRA) to model this question, and our proposed method can reconstruct arbitrary filters theoretically. We also conduct node classification experiments on real-world graph benchmarks and achieve superior performance on most datasets. Our code is available at https://github.com/Bufordyang/WaveNet

Video inpainting aims to fill in the missing regions of the video frames with plausible content. Benefiting from the outstanding long-range modeling capacity, the transformer-based models have achieved unprecedented performance regarding inpainting quality. Essentially, coherent contents from all the frames along both spatial and temporal dimensions are concerned by a patch-wise attention module, and then the missing contents are generated based on the attention-weighted summation. In this way, attention retrieval accuracy has become the main bottleneck to improve the video inpainting performance, where the factors affecting attention calculation should be explored to maximize the advantages of transformer. Towards this end, in this paper, we theoretically certificate that noise is the culprit that entangles the process of attention calculation. Meanwhile, we propose a novel wavelet transformer network with noise robustness for video inpainting, named WaveFormer. Unlike existing transformer-based methods that utilize the whole embeddings to calculate the attention, our WaveFormer first separates the noise existing in the embedding into high-frequency components by introducing the Discrete Wavelet Transform (DWT), and then adopts clean low-frequency components to calculate the attention. In this way, the impact of noise on attention computation can be greatly mitigated and the missing content regarding different frequencies can be generated by sharing the calculated attention. Extensive experiments validate the superior performance of our method over state-of-the-art baselines both qualitatively and quantitatively.

To mitigate potential risks associated with language models (LMs), recent AI detection research proposes incorporating watermarks into machine-generated text through random vocabulary restrictions and utilizing this information for detection. In this paper, we show that watermarking algorithms designed for LMs cannot be seamlessly applied to conditional text generation (CTG) tasks without a notable decline in downstream task performance. To address this issue, we introduce a simple yet effective semantic-aware watermarking algorithm that considers the characteristics of conditional text generation with the input context. Compared to the baseline watermarks, our proposed watermark yields significant improvements in both automatic and human evaluations across various text generation models, including BART and Flan-T5, for CTG tasks such as summarization and data-to-text generation. Meanwhile, it maintains detection ability with higher z-scores but lower AUC scores, suggesting the presence of a detection paradox that poses additional challenges for watermarking CTG.

Federated learning encounters substantial challenges with heterogeneous data, leading to performance degradation and convergence issues. While considerable progress has been achieved in mitigating such an impact, the reliability aspect of federated models has been largely disregarded. In this study, we conduct extensive experiments to investigate the reliability of both generic and personalized federated models. Our exploration uncovers a significant finding: federated models exhibit unreliability when faced with heterogeneous data, demonstrating poor calibration on in-distribution test data and low uncertainty levels on out-of-distribution data. This unreliability is primarily attributed to the presence of biased projection heads, which introduce miscalibration into the federated models. Inspired by this observation, we propose the "Assembled Projection Heads" (APH) method for enhancing the reliability of federated models. By treating the existing projection head parameters as priors, APH randomly samples multiple initialized parameters of projection heads from the prior and further performs targeted fine-tuning on locally available data under varying learning rates. Such a head ensemble introduces parameter diversity into the deterministic model, eliminating the bias and producing reliable predictions via head averaging. We evaluate the effectiveness of the proposed APH method across three prominent federated benchmarks. Experimental results validate the efficacy of APH in model calibration and uncertainty estimation. Notably, APH can be seamlessly integrated into various federated approaches but only requires less than 30% additional computation cost for 100x inferences within large models.

Differential privacy (DP) has achieved remarkable results in the field of privacy-preserving machine learning. However, existing DP frameworks do not satisfy all the conditions for becoming metrics, which prevents them from deriving better basic private properties and leads to exaggerated values on privacy budgets. We propose Wasserstein differential privacy (WDP), an alternative DP framework to measure the risk of privacy leakage, which satisfies the properties of symmetry and triangle inequality. We show and prove that WDP has 13 excellent properties, which can be theoretical supports for the better performance of WDP than other DP frameworks. In addition, we derive a general privacy accounting method called Wasserstein accountant, which enables WDP to be applied in stochastic gradient descent (SGD) scenarios containing subsampling. Experiments on basic mechanisms, compositions and deep learning show that the privacy budgets obtained by Wasserstein accountant are relatively stable and less influenced by order. Moreover, the overestimation on privacy budgets can be effectively alleviated. The code is available at https://github.com/Hifipsysta/WDP.

Current weakly-supervised semantic segmentation (WSSS) techniques concentrate on enhancing class activation maps (CAMs) with image-level annotations. Yet, the emphasis on producing these pseudo-labels often overshadows the pivotal role of training the segmentation model itself. This paper underscores the significant influence of noisy pseudo-labels on segmentation network performance, particularly in boundary region. To address above issues, we introduce a novel paradigm: Weak to Partial Supervision (W2P). At its core, W2P categorizes the pseudo-labels from WSSS into two unique supervisions: trustworthy clean labels and uncertain noisy labels. Next, our proposed partially-supervised framework adeptly employs these clean labels to rectify the noisy ones, thereby promoting the continuous enhancement of the segmentation model. To further optimize boundary segmentation, we incorporate a noise detection mechanism that specifically preserves boundary regions while eliminating noise. During the noise refinement phase, we adopt a boundary-conscious noise correction technique to extract comprehensive boundaries from noisy areas. Furthermore, we devise a boundary generation approach that assists in predicting intricate boundary zones. Evaluations on the PASCAL VOC 2012 and MS COCO 2014 datasets confirm our method's impressive segmentation capabilities across various pseudo-labels.

Efficient representation of point clouds is fundamental for LiDAR-based 3D object detection. While recent grid-based detectors often encode point clouds into either voxels or pillars, the distinctions between these approaches remain underexplored. In this paper, we quantify the differences between the current encoding paradigms and highlight the limited vertical learning within. To tackle these limitations, we propose a hybrid detection framework named Voxel-Pillar Fusion (VPF), which synergistically combines the unique strengths of both voxels and pillars. To be concrete, we first develop a sparse voxel-pillar encoder that encodes point clouds into voxel and pillar features through 3D and 2D sparse convolutions respectively, and then introduce the Sparse Fusion Layer (SFL), facilitating bidirectional interaction between sparse voxel and pillar features. Our computationally efficient, fully sparse method can be seamlessly integrated into both dense and sparse detectors. Leveraging this powerful yet straightforward representation, VPF delivers competitive performance, achieving real-time inference speeds on the nuScenes and Waymo Open Dataset.

Composite images (CIs) typically combine various elements from different scenes, views, and styles, which are a very important information carrier in the era of mixed media such as virtual reality, mixed reality, metaverse, etc. However, the complexity of CI content presents a significant challenge for subsequent visual perception modeling and compression. In addition, the lack of benchmark CI databases also hinders the use of recent advanced data-driven methods. To address these challenges, we first establish one of the earliest visual redundancy prediction (VRP) databases for CIs. Moreover, we propose a multi-visual effect (MVE)-driven incremental learning method that combines the strengths of hand-crafted and data-driven approaches to achieve more accurate VRP modeling. Specifically, we design special incremental rules to learn the visual knowledge flow of MVE. To effectively capture the associated features of MVE, we further develop a three-stage incremental learning approach for VRP based on an encoder-decoder network. Extensive experimental results validate the superiority of the proposed method in terms of subjective, objective, and compression experiments.

Visual Language is a multitasking on-line system focusing on e-commerce, which involves in generating accurate product descriptions for sellers and providing convenient product retrieval service for customers. To achieve this goal, the system adopts image description technology and multi-modal retrieval technology. By utilizing cross-modal generation technique, we could help sellers on rapid uploading products and customers on rapid retrieval, which could improve the experience of both sellers and customers.