Vehicle-to-everything (V2X) collaborative perception has recently gained increasing attention in autonomous driving due to its ability to enhance scene understanding by integrating information from other collaborators, e.g. vehicles or infrastructure. Existing algorithms usually share deep features to achieve a trade-off between accuracy and bandwidth. However, most of these methods require joint training of all agents, which results in privacy leakage and is impractical and unacceptable in the real world. Sharing prediction results seems to be a direct solution, but its performance is suboptimal and sensitive to localization noise and communication delay. In this paper, we propose a privacy-preserving collaborative perception framework, where each agent is separately trained with its own dataset and the ego vehicle needs to integrate with completely unknown collaborators. Specifically, we propose MSD, a multi-scale feature fusion method combined with deformable attention, to better fuse features of different agents. We also propose a plug-in domain adapter to align the features from unknown collaborators to ego-domain. Extensive experiments on the challenging DAIR-V2X and V2V4Real demonstrate that: 1) MSD achieves remarkable performance, outperforming others by at least 2.8% and 6.7% in AP0.7 on DAIR-V2X and V2V4Real, respectively; 2) After domain adaptation, it significantly outperforms the No Fusion, Late Fusion scenarios and can approach or even surpass the performance of joint training. We truly achieves privacy-preserving collaboration, providing a new paradigm for the study of collaborative perception, which is crucial for practical applications.
论文检索
输入标题、作者或关键词，从 11,272 篇学术成果中精准定位
Existing RGBT tracking methods often design various interaction models to perform cross-modal fusion of each layer, but can not execute the feature interactions among all layers, which plays a critical role in robust multimodal representation, due to large computational burden. To address this issue, this paper presents a novel All-layer multimodal Interaction Network, named AINet, which performs efficient and effective feature interactions of all modalities and layers in a progressive fusion Mamba, for robust RGBT tracking. Even though modality features in different layers are known to contain different cues, it is always challenging to build multimodal interactions in each layer due to struggling in balancing interaction capabilities and efficiency. Meanwhile, considering that the feature discrepancy between RGB and thermal modalities reflects their complementary information to some extent, we design a Difference-based Fusion Mamba (DFM) to achieve enhanced fusion of different modalities with linear complexity. When interacting with features from all layers, a huge number of token sequences (3840 tokens in this work) are involved and the computational burden is thus large. To handle this problem, we design an Order-dynamic Fusion Mamba (OFM) to execute efficient and effective feature interactions of all layers by dynamically adjusting the scan order of different layers in Mamba. Extensive experiments on four public RGBT tracking datasets show that AINet achieves leading performance against existing state-of-the-art methods. We will release the code upon acceptance of the paper.
Enhancing Low-Light Images: A Synthetic Data Perspective on Practical and Generalizable Solutions
PDF ↗Recently, deep neural networks (DNNs) have emerged as the leading approach for low-light image enhancement (LLIE). However, training these models generally requires large-scale paired datasets, which are challenging to obtain due to the labor-intensive and time-consuming nature of real-world data collection. To alleviate this issue, synthetic data are often combined with real-captured data for training. However, most existing low-light image synthesis methods are simply performed in the sRGB domain using Gamma correction or manual adjustments via Lightroom, which fail to incorporate the physical imaging prior through the image signal processing (ISP) pipeline and thus result in limited dataset size and degradation space. Consequently, LLIE methods trained on such data often exhibit some drawbacks in the results, such as inaccurate white balance and abnormal enhancement artifacts, which limit their practicality and generalizability. In this paper, we propose a practical low-light image synthesis pipeline capable of generating unlimited paired training data. Our pipeline starts with a reverse ISP model that converts sRGB images back to the unprocessed RAW domain, where we then simulate low-light degradation, noise degradation, and white balance adjustments. Finally, the degraded RAW images are processed through a forward ISP model to produce low-light sRGB images. The pipeline further employs multiple tone mapping curves and color correction matrices (CCMs) to expand the degradation space. Hence, trained with our proposed synthetic data, existing state-of-the-art (SOTA) LLIE deep models are expected to improve their performance. Extensive experiments across various datasets demonstrate that our synthetic data can indeed effectively enhance existing LLIE deep models, improving both their practicality and generalizability.
The Segment Anything Model (SAM) is a widely used vision foundation model with diverse applications, including image segmentation, detection, and tracking. Given SAM's wide applications, understanding its robustness against adversarial attacks is crucial for real-world deployment. However, research on SAM's robustness is still in its early stages. Existing attacks often overlook the role of prompts in evaluating SAM's robustness, and there has been insufficient exploration of defense methods to balance the robustness and accuracy. To address these gaps, this paper proposes an adversarial robustness framework designed to evaluate and enhance the robustness of SAM. Specifically, we introduce a cross-prompt attack method to enhance the attack transferability across different prompt types. Besides attacking, we propose a few-parameter adaptation strategy to defend SAM against various adversarial attacks. To balance robustness and accuracy, we use the singular value decomposition (SVD) to constrain the space of trainable parameters, where only singular values are adaptable. Experiments demonstrate that our cross-prompt attack method outperforms previous approaches in terms of attack success rate on both SAM and SAM 2. By adapting only 512 parameters, we achieve at least a 15% improvement in mean intersection over union (mIoU) against various adversarial attacks. Compared to previous defense methods, our approach enhances the robustness of SAM while maximally maintaining its original performance.
Edge labels are typically at various granularity levels owing to the varying preferences of annotators, thus handling the subjectivity of per-pixel labels has been a focal point for edge detection. Previous methods often employ a simple voting strategy to diminish such label uncertainty or impose a strong assumption of labels with a pre-defined distribution, e.g., Gaussian. In this work, we unveil that the segment anything model (SAM) provides strong prior knowledge to model the uncertainty in edge labels. Our key insight is that the intermediate SAM features inherently correspond to object edges at various granularities, which reflects different edge options due to uncertainty. Therefore, we attempt to align uncertainty with granularity by regressing intermediate SAM features from different layers to object edges at multi-granularity levels. In doing so, the model can fully and explicitly explore diverse ``uncertainties'' in a data-driven fashion. Specifically, we inject a lightweight module (~ 1.5% additional parameters) into the frozen SAM to progressively fuse and adapt its intermediate features to estimate edges from coarse to fine. It is crucial to normalize the granularity level of human edge labels to match their innate uncertainty. For this, we simply perform linear blending to the real edge labels at hand to create pseudo labels with varying granularities. Consequently, our uncertainty-aligned edge detector can flexibly produce edges at any desired granularity (including an optimal one). Thanks to SAM, our model uniquely demonstrates strong generalizability for cross-dataset edge detection. Extensive experimental results on BSDS500, Muticue and NYUDv2 validate our model's superiority.
Formal verification provides critical security assurances for neural networks, yet its practical application suffers from the long verification time. This work introduces a novel method for training verification-friendly neural networks, which are robust, easy to verify, and relatively accurate. Our method integrates neuron behavior consistency into the training process, making neuron activation states remain consistent across different inputs within a local neighborhood. This reduces the number of unstable neurons and tightens the bounds of neurons thereby enhancing the network's verifiability. We evaluated our method using the MNIST, Fashion-MNIST, and CIFAR-10 datasets with various network architectures. The experimental results demonstrate that networks trained using our method are verification-friendly across different radii and architectures, whereas other tools fail to maintain verifiability as the radius increases. Additionally, we show that our method can be combined with existing approaches to further improve the verifiability of networks.
Advancing Comprehensive Aesthetic Insight with Multi-Scale Text-Guided Self-Supervised Learning
PDF ↗Image Aesthetic Assessment (IAA) is a vital and intricate task that entails analyzing and assessing an image's aesthetic values, and identifying its highlights and areas for improvement. Traditional methods of IAA often concentrate on a single aesthetic task and suffer from inadequate labeled datasets, thus impairing in-depth aesthetic comprehension. Despite efforts to overcome this challenge through the application of Multi-modal Large Language Models (MLLMs), such models remain underdeveloped for IAA purposes. To address this, we propose a comprehensive aesthetic MLLM capable of nuanced aesthetic insight. Central to our approach is an innovative multi-scale text-guided self-supervised learning technique. This technique features a multi-scale feature alignment module and capitalizes on a wealth of unlabeled data in a self-supervised manner to structurally and functionally enhance aesthetic ability. The empirical evidence indicates that accompanied with extensive instruct-tuning, our model sets new state-of-the-art benchmarks across multiple tasks, including aesthetic scoring, aesthetic commenting, and personalized image aesthetic assessment. Remarkably, it also demonstrates zero-shot learning capabilities in the emerging task of aesthetic suggesting. Furthermore, for personalized image aesthetic assessment, we harness the potential of in-context learning and showcase its inherent advantages.
Sign languages, used by around 70 million Deaf individuals globally, are visual languages that convey visual and contextual information. Current methods in vision-based sign language recognition (SLR) and translation (SLT) struggle with dialogue scenes due to limited dataset diversity and the neglect of contextually relevant information. To address these challenges, we introduce SCOPE (Sign language COntextual Processing with Embedding from LLMs), a novel context-aware vision-based SLR and SLT framework. For SLR, we utilize dialogue contexts through a multi-modal encoder to enhance gloss-level recognition. For subsequent SLT, we further fine-tune a Large Language Model (LLM) by incorporating prior conversational context. We also contribute a new sign language dataset that contains 72 hours of Chinese sign language videos in contextual dialogues across various scenarios. Experimental results demonstrate that our SCOPE framework achieves state-of-the-art performance on multiple datasets, including Phoenix-2014T, CSL-Daily, and our SCOPE dataset. Moreover, surveys conducted with participants from the Deaf community further validate the robustness and effectiveness of our approach in real-world applications. Both our dataset and code will be open-sourced to facilitate further research.
Deciphering visual content from fMRI sheds light on the human vision system, but data scarcity and noise limit brain decoding model performance. Traditional approaches rely on subject-specific models, which are sensitive to training sample size. In this paper, we address data scarcity by proposing shallow subject-specific adapters to map cross-subject fMRI data into unified representations. A shared deep decoding model then decodes these features into the target feature space. We use both visual and textual supervision for multi-modal brain decoding and integrate high-level perception decoding with pixel-wise reconstruction guided by high-level perceptions. Our extensive experiments reveal several interesting insights: 1) Training with cross-subject fMRI benefits both high-level and low-level decoding models; 2) Merging high-level and low-level information improves reconstruction performance at both levels; 3) Transfer learning is effective for new subjects with limited training data by training new adapters; 4) Decoders trained on visually-elicited brain activity can generalize to decode imagery-induced activity, though with reduced performance.
Visual Question Answering (VQA) has garnered significant attention as a crucial link between vision and language, aimed at generating accurate responses to visual queries. However, current VQA models still struggle with the challenges of minority class collapse and spurious semantic correlations posed by language bias and imbalanced distributions. To address these challenges, this paper proposes a novel Prompt-Driven Geometric Harmonization (PDGH) paradigm, which integrates both geometric structure and information entropy principles to enhance the ability of VQA models to generalize effectively across diverse scenarios. Specifically, our PDGH approach is meticulously designed to generate image-generated prompts that are guided by specific question cues, facilitating a more accurate and context-aware understanding of the visual content. Moreover, we project the prompt-visual-question and visual-question joint representations into a unified hypersphere space, applying feature weight self-orthogonality and prompt-information entropy correction constraints to optimize the margin, further alleviating minority class collapse and correcting language bias. To maintain the geometric integrity of the representation space, we introduce multi-space geometric contrast constraints to minimize the impact of spurious priors introduced during training. Finally, a semantic matrix is constructed for the coordinated joint representation to ensure that the learned instances are semantically consistent and improve reasoning ability. Extensive experiments on various general and medical VQA datasets demonstrate the consistent superiority of our PDGH approach over existing state-of-the-art baselines.
DreamUHD: Frequency Enhanced Variational Autoencoder for Ultra-High-Definition Image Restoration
PDF ↗Existing ultra-high-definition (UHD) image restoration methods often struggle with consistency due to downsampling. We aim to address these challenges by leveraging the powerful latent space representation and reconstruction capabilities of Variational Autoencoders (VAE). However, applying VAE to UHD image restoration presents challenges: 1) High-performing VAEs have large parameter sizes, leading to significant carbon footprints; 2) The self-reconstruction property of VAE hinders bridging the domain gap between clean and degraded images; 3) Latent encoding in VAE can lose high-frequency information, compromising image detail. To overcome these challenges, we propose a frequency enhanced VAE UHD image restoration framework by integrating frequency priors. First, we design the Fourier-based lightweight frequency learning within the VAE to improve parameter efficiency. Then, we introduce a wavelet-based adapter that extracts multi-scale image information and employs frequency-aware adaptive modulation to bridge the domain gap by integrating degraded image data into the pre-trained VAE. Additionally, the adapter injects high-frequency information into the VAE decoder, enhancing detail in the restored images. In this way, our method effectively combines the powerful latent space representation with frequency priors to enhance UHD image restoration. Extensive experiments on various UHD image restoration tasks show that our method surpasses state-of-the-art methods both qualitatively and quantitatively.
Text-based person retrieval (TPR) has gained significant attention as a fine-grained and challenging task that closely aligns with practical applications. Tailoring CLIP to person domain is now a emerging research topic due to the abundant knowledge of vision-language pretraining, but challenges still remain during fine-tuning: (i) Previous full-model fine-tuning in TPR is computationally expensive and prone to overfitting.(ii) Existing parameter-efficient transfer learning (PETL) for TPR lacks of fine-grained feature extraction. To address these issues, we propose Domain-Aware Mixture-of-Adapters (DM-Adapter), which unifies Mixture-of-Experts (MOE) and PETL to enhance fine-grained feature representations while maintaining efficiency. Specifically, Sparse Mixture-of-Adapters is designed in parallel to MLP layers in both vision and language branches, where different experts specialize in distinct aspects of person knowledge to handle features more finely. To promote the router to exploit domain information effectively and alleviate the routing imbalance, Domain-Aware Router is then developed by building a novel gating function and injecting learnable domain-aware prompts. Extensive experiments show that our DM-Adapter achieves state-of-the-art performance, outperforming previous methods by a significant margin.
CLIP-PCQA: Exploring Subjective-Aligned Vision-Language Modeling for Point Cloud Quality Assessment
PDF ↗In recent years, No-Reference Point Cloud Quality Assessment (NR-PCQA) research has achieved significant progress. However, existing methods mostly seek a direct mapping function from visual data to the Mean Opinion Score (MOS), which is contradictory to the mechanism of practical subjective evaluation. To address this, we propose a novel language-driven PCQA method named CLIP-PCQA. Considering that human beings prefer to describe visual quality using discrete quality descriptions (e.g., "excellent" and "poor") rather than specific scores, we adopt a retrieval-based mapping strategy to simulate the process of subjective assessment. More specifically, based on the philosophy of CLIP, we calculate the cosine similarity between the visual features and multiple textual features corresponding to different quality descriptions, in which process an effective contrastive loss and learnable prompts are introduced to enhance the feature extraction. Meanwhile, given the personal limitations and bias in subjective experiments, we further covert the feature similarities into probabilities and consider the Opinion Score Distribution (OSD) rather than a single MOS as the final target. Experimental results show that our CLIP-PCQA outperforms other State-Of-The-Art (SOTA) approaches.
Image forgeries can entirely change the semantic information of an image, and can be used for unscrupulous purposes. In this paper, we propose a novel image forgery localization network named as MUN, which consists of an M^3 encoder and a UN decoder. Firstly, the M^3 encoder is constructed based on a Multi-scale Max-pooling query module to extract Multi-clue forged features. Noiseprint++ is adopted to assist the RGB clue, and its deployment methodology is discussed. A Multi-scale Max-pooling Query (MMQ) module is proposed to integrate RGB and noise features. Secondly, a novel UN decoder is proposed to extract hierarchical features from both top-down and bottom-up directions, reconstructing both high-level and low-level features at the same time. Thirdly, we formulate an IoU-recalibrated Dynamic Cross-Entropy (IoUDCE) loss to dynamically adjust the weights on forged regions according to IoU which can adaptively balance the influence of authentic and forged regions. Last but not least, we propose a data augmentation method, i.e., Deviation Noise Augmentation (DNA), which acquires accessible prior knowledge of RGB distribution to improve the generalization ability. Extensive experiments on publicly available datasets show that MUN outperforms the state-of-the-art works.
Learning visual semantic similarity is a critical challenge in bridging the gap between images and texts. However, there exist inherent variations between vision and language data, such as information density, i.e., images can contain textual information from multiple different views, which makes it difficult to compute the similarity between these two modalities accurately and efficiently. In this paper, we propose a novel framework called Asymmetric Visual Semantic Embedding (AVSE) to dynamically select features from various regions of images tailored to different textual inputs for similarity calculation. To capture information from different views in the image, we design a radial bias sampling module to sample image patches and obtain image features from various views, Furthermore, AVSE introduces a novel module for efficient computation of visual semantic similarity between asymmetric image and text embeddings. Central to this module is the presumption of foundational semantic units within the embeddings, denoted as ``meta-semantic embeddings." It segments all embeddings into meta-semantic embeddings with the same dimension and calculates visual semantic similarity by finding the optimal match of meta-semantic embeddings of two modalities. Our proposed AVSE model is extensively evaluated on the large-scale MS-COCO and Flickr30K datasets, demonstrating its superiority over recent state-of-the-art methods.
Learning Dynamic Similarity by Bidirectional Hierarchical Sliding Semantic Probe for Efficient Text Video Retrieval
PDF ↗Text-video retrieval is a foundation task in multi-modal research which aims to align texts and videos in the embedding space. The key challenge is to learn the similarity between videos and texts. A conventional approach involves directly aligning video-text pairs using cosine similarity. However, due to the disparity in the information conveyed by videos and texts, i.e., a single video can be described from multiple perspectives, the retrieval accuracy is suboptimal. An alternative approach employs cross-modal interaction to enable videos to dynamically acquire distinct features from various texts, thus facilitating similarity calculations. Nevertheless, this solution incurs a computational complexity of O(n^2) during retrieval. To this end, this paper proposes a novel method called Bidirectional Hierarchical Sliding Semantic Probe (BiHSSP), which calculates dynamic similarity between videos and texts with O(n) complexity during retrieval. We introduce a hierarchical semantic probe module that learns semantic probes at different scales for both video and text features. Semantic probe involves a sliding calculation of the cross-correlation between semantic probes at different scales and embeddings from another modality, allowing for dynamic similarity computation between video and text descriptions from various perspectives. Specifically, for text descriptions from different angles, we calculate the similarity at different locations within the video features and vice versa. This approach preserves the complete information of the video while addressing the issue of unequal information between video and text without requiring cross-modal interaction. Additionally, our method can function as a plug-and-play module across various methods, thereby enhancing the corresponding performance. Experimental results demonstrate that our BiHSSP significantly outperforms the baseline.
Recent advances in vision-language pre-training have significantly enhanced the model capabilities on grounded object detection. However, these studies often pre-train with coarse-grained text prompts, such as plain category names and brief grounded phrases. This limitation curtails the model's capacity for fine-grained linguistic comprehension and leads to a significant decline in performance when faced with detailed descriptions or contextual information. To tackle these problems, we develop DoGA: Detect objects with Grouped Attributes, which employs commonly apparent attributes to bridge different granular semantics and uses specific attributes to identify the object discrepancy. Our DoGA incorporates three principle components: 1) Generation of attribute-based prompts, consisting of linguistic definitions enriched with common-sense visible attributes and hard negative notations deriving from the image-specific attribute features; 2) Paralleled entity fusion and optimization, designed to manage long attribute-based descriptions and negative concepts efficiently; and 3) Prompt-wise grouped training to accommodate model to perform many-to-many assignments, facilitating simultaneous training and inferring with multiple attribute-based synonyms. Extensive experiments demonstrate that training with synonymous attribute-based prompts allows DoGA to generalize multi-granular prompts and surpass previous state-of-the-art approaches, yielding 50.2 on the COCO and 38.0 on the LVIS benchmarks under the zero-short setting. We will make our code publicly available upon acceptance.
Training-free open-vocabulary semantic segmentation aims to explore the potential of frozen vision-language models (VLM) for segmentation tasks. Recent works reform the inference process of CLIP and utilize the features from the final layer to reconstruct dense representations for segmentation, demonstrating promising performance. However, the final layer tends to prioritize global components over local representations, leading to suboptimal robustness and effectiveness of existing methods. In this paper, we propose CLIPSeg, a novel training-free framework that fully exploits the diverse knowledge across layers in CLIP for dense predictions. Our study unveils two key discoveries: Firstly, the features in the middle layers exhibit high locality awareness and feature coherence compared to the final layer, based on which we propose the coherence enhanced residual attention module that generates semantic-aware attention. Secondly, despite not being directly aligned with the text, the deep layers capture valid local semantics that complement those in the final layer. Leveraging this insight, we introduce the deep semantic integration module to boost the patch semantics in the final block. Experiments conducted on 9 segmentation benchmarks with various CLIP models demonstrate that CLIPSeg consistently outperforms all training-free methods by substantial margins, e.g., a 7.8 % improvement in average mIoU for CLIP with a ViT-L backbone, and competes with learning-based counterparts in generalizing to novel concepts in an efficient way.
Knowledge Distillation (KD) is a promising approach for unsupervised Anomaly Detection (AD). However, the student network's over-generalization often diminishes the crucial representation differences between teacher and student in anomalous regions, leading to detection failures. To address this problem, the widely accepted Reverse Distillation (RD) paradigm designs the asymmetry teacher and student network, using an encoder as teacher and a decoder as student. Yet, the design of RD does not ensure that the teacher encoder effectively distinguishes between normal and abnormal features or that the student decoder generates anomaly-free features. Additionally, the absence of skip connections results in a loss of fine details during feature reconstruction. To address these issues, we propose RD with Expert, which introduces a novel Expert-Teacher-Student network for simultaneous distillation of both the teacher encoder and student decoder. The added expert network enhances the student's ability to generate normal features and optimizes the teacher's differentiation between normal and abnormal features, reducing missed detections. Additionally, Guided Information Injection is designed to filter and transfer features from teacher to student, improving detail reconstruction and minimizing false positives. Experiments on several benchmarks prove that our method outperforms existing unsupervised AD methods under RD paradigm, fully unlocking RD’s potential.
Semi-supervised hashing has shown promising efficacy in large-scale image retrieval, which learns similarity-preserving codes from both labeled and unlabeled data. To enable the use of advanced supervised hashing techniques, pseudo labels are widely applied. However, existing methods typically suffer from a biased learning issue due to pseudo label noise, which can be further aggravated during optimization. Although such a bias can adversely affect hashing accuracy, it has not been investigated sufficiently. In view of this, we present a comprehensive discussion on potential causes of biases, involving processes of pseudo-labeling, hash learning and optimization. Accordingly, a novel Generalized Debiased Semi-supervised Hashing (GDSH) method is proposed as a unified solution to mitigate the biases. Specifically, reliable pseudo labels are first predicted via a robust label completion strategy. Secondly, a debiased hash learning module is designed by combining label denoising and similarity updating. This can not only refine the supervision, but also obtain hash codes that are semantically debiased in both category and sample levels. Finally, a discrete semi-supervised hashing algorithm is proposed to alleviate the bias arising from optimization. Experimental results on three single-label and three multi-label image benchmarks demonstrate that GDSH remarkably outperforms the state-of-the-arts in different semi-supervised settings.