Search

Pre-training for Reinforcement Learning (RL) with purely video data is a valuable yet challenging problem. Although in-the-wild videos are readily available and inhere a vast amount of prior world knowledge, the absence of action annotations and the common domain gap with downstream tasks hinder utilizing videos for RL pre-training. To address the challenge of pre-training with videos, we propose Pre-trained Visual Dynamics Representations (PVDR) to bridge the domain gap between videos and downstream tasks for efficient policy learning. By adopting video prediction as a pre-training task, we use a Transformer-based Conditional Variational Autoencoder (CVAE) to learn visual dynamics representations. The pre-trained visual dynamics representations capture the visual dynamics prior knowledge in the videos. This abstract prior knowledge can be readily adapted to downstream tasks and aligned with executable actions through online adaptation. We conduct experiments on a series of robotics visual control tasks and verify that PVDR is an effective form for pre-training with videos to promote policy learning., Comment: ECCV 2024

The transferable belief model, as a semantic interpretation of Dempster-Shafer theory, enables agents to perform reasoning and decision making in imprecise and incomplete environments. The model offers distinct semantics for handling unreliable testimonies, allowing for a more reasonable and general process of belief transfer compared to the Bayesian approach. However, because both the belief masses and the structure of focal sets must be considered when updating belief functions-leading to extra computational complexity during reasoning-the transferable belief model has gradually lost favor among researchers in recent developments. In this paper, we implement the transferable belief model on quantum circuits and demonstrate that belief functions offer a more concise and effective alternative to Bayesian approaches within the quantum computing framework. Furthermore, leveraging the unique characteristics of quantum computing, we propose several novel belief transfer approaches. More broadly, this paper introduces a new perspective on basic information representation for quantum AI models, suggesting that belief functions are more suitable than Bayesian approach for handling uncertainty on quantum circuits.

Diffusion models have made compelling progress on facilitating high-throughput daily production. Nevertheless, the appealing customized requirements are remain suffered from instance-level finetuning for authentic fidelity. Prior zero-shot customization works achieve the semantic consistence through the condensed injection of identity features, while addressing detailed low-level signatures through complex model configurations and subject-specific fabrications, which significantly break the statistical coherence within the overall system and limit the applicability across various scenarios. To facilitate the generic signature concentration with rectified efficiency, we present \textbf{AnyLogo}, a zero-shot region customizer with remarkable detail consistency, building upon the symbiotic diffusion system with eliminated cumbersome designs. Streamlined as vanilla image generation, we discern that the rigorous signature extraction and creative content generation are promisingly compatible and can be systematically recycled within a single denoising model. In place of the external configurations, the gemini status of the denoising model promote the reinforced subject transmission efficiency and disentangled semantic-signature space with continuous signature decoration. Moreover, the sparse recycling paradigm is adopted to prevent the duplicated risk with compressed transmission quota for diversified signature stimulation. Extensive experiments on constructed logo-level benchmarks demonstrate the effectiveness and practicability of our methods., Comment: 13 pages, 12 figures

Recently, 3D Gaussian Splatting (3DGS) has garnered attention for its high fidelity and real-time rendering. However, adapting 3DGS to different camera models, particularly fisheye lenses, poses challenges due to the unique 3D to 2D projection calculation. Additionally, there are inefficiencies in the tile-based splatting, especially for the extreme curvature and wide field of view of fisheye lenses, which are crucial for its broader real-life applications. To tackle these challenges, we introduce Fisheye-GS.This innovative method recalculates the projection transformation and its gradients for fisheye cameras. Our approach can be seamlessly integrated as a module into other efficient 3D rendering methods, emphasizing its extensibility, lightweight nature, and modular design. Since we only modified the projection component, it can also be easily adapted for use with different camera models. Compared to methods that train after undistortion, our approach demonstrates a clear improvement in visual quality.

Scene observation from multiple perspectives would bring a more comprehensive visual experience. However, in the context of acquiring multiple views in the dark, the highly correlated views are seriously alienated, making it challenging to improve scene understanding with auxiliary views. Recent single image-based enhancement methods may not be able to provide consistently desirable restoration performance for all views due to the ignorance of potential feature correspondence among different views. To alleviate this issue, we make the first attempt to investigate multi-view low-light image enhancement. First, we construct a new dataset called Multi-View Low-light Triplets (MVLT), including 1,860 pairs of triple images with large illumination ranges and wide noise distribution. Each triplet is equipped with three different viewpoints towards the same scene. Second, we propose a deep multi-view enhancement framework based on the Recurrent Collaborative Network (RCNet). Specifically, in order to benefit from similar texture correspondence across different views, we design the recurrent feature enhancement, alignment and fusion (ReEAF) module, in which intra-view feature enhancement (Intra-view EN) followed by inter-view feature alignment and fusion (Inter-view AF) is performed to model the intra-view and inter-view feature propagation sequentially via multi-view collaboration. In addition, two different modules from enhancement to alignment (E2A) and from alignment to enhancement (A2E) are developed to enable the interactions between Intra-view EN and Inter-view AF, which explicitly utilize attentive feature weighting and sampling for enhancement and alignment, respectively. Experimental results demonstrate that our RCNet significantly outperforms other state-of-the-art methods. All of our dataset, code, and model will be available at https://github.com/hluo29/RCNet., Comment: 14 Pages, 10 Figures, Under Review

In this paper, based a novel primal-dual dynamical model with adaptive scaling parameters and Bregman divergences, we propose new accelerated primal-dual proximal gradient splitting methods for solving bilinear saddle-point problems with provable optimal nonergodic convergence rates. For the first, using the spectral analysis, we show that a naive extension of acceleration model for unconstrained optimization problems to a quadratic game is unstable. Motivated by this, we present an accelerated primal-dual hybrid gradient (APDHG) flow which combines acceleration with careful velocity correction. To work with non-Euclidean distances, we also equip our APDHG model with general Bregman divergences and prove the exponential decay of a Lyapunov function. Then, new primal-dual splitting methods are developed based on proper semi-implicit Euler schemes of the continuous model, and the theoretical convergence rates are nonergodic and optimal with respect to the matrix norms,\, Lipschitz constants and convexity parameters. Thanks to the primal and dual scaling parameters, both the algorithm designing and convergence analysis cover automatically the convex and (partially) strongly convex objectives. Moreover, the use of Bregman divergences not only unifies the standard Euclidean distances and general cases in an elegant way, but also makes our methods more flexible and adaptive to problem-dependent metrics.

Compared with the remarkable progress made in parallel numerical solvers of partial differential equations,the development of algorithms for generating unstructured triangular/tetrahedral meshes has been relatively sluggish. In this paper, we propose a novel, consistent parallel advancing front technique (CPAFT) by combining the advancing front technique, the domain decomposition method based on space-filling curves, the distributed forest-of-overlapping-trees approach, and the consistent parallel maximal independent set algorithm. The newly proposed CPAFT algorithm can mathematically ensure that the generated unstructured triangular/tetrahedral meshes are independent of the number of processors and the implementation of domain decomposition. Several numerical tests are conducted to validate the parallel consistency and outstanding parallel efficiency of the proposed algorithm, which scales effectively up to two thousand processors. This is, as far as we know, the first parallel unstructured triangular/tetrahedral mesh generator with scalability to O(1,000) CPU processors.

In recent years, various intelligent autonomous robots have begun to appear in daily life and production. Desktop-level robots are characterized by their flexible deployment, rapid response, and suitability for light workload environments. In order to meet the current societal demand for service robot technology, this study proposes using a miniaturized desktop-level robot (by ROS) as a carrier, locally deploying a natural language model (NLP-BERT), and integrating visual recognition (CV-YOLO) and speech recognition technology (ASR-Whisper) as inputs to achieve autonomous decision-making and rational action by the desktop robot. Three comprehensive experiments were designed to validate the robotic arm, and the results demonstrate excellent performance using this approach across all three experiments. In Task 1, the execution rates for speech recognition and action performance were 92.6% and 84.3%, respectively. In Task 2, the highest execution rates under the given conditions reached 92.1% and 84.6%, while in Task 3, the highest execution rates were 95.2% and 80.8%, respectively. Therefore, it can be concluded that the proposed solution integrating ASR, NLP, and other technologies on edge devices is feasible and provides a technical and engineering foundation for realizing multimodal desktop-level robots.

We give a continuous perspective on the Inertial Corrected Primal-Dual Proximal Splitting (IC-PDPS) proposed by Valkonen ({\it SIAM J. Optim.}, 30(2): 1391--1420, 2020) for solving saddle-point problems. The algorithm possesses nonergodic convergence rate and admits a tight preconditioned proximal point formulation which involves both inertia and additional correction. Based on new understandings on the relation between the discrete step size and rescaling effect, we rebuild IC-PDPS as a semi-implicit Euler scheme with respect to its iterative sequences and integrated parameters. This leads to two novel second-order ordinary differential equation (ODE) models that are equivalent under proper time transformation, and also provides an alternative interpretation from the continuous point of view. Besides, we present the convergence analysis of the Lagrangian gap along the continuous trajectory by using proper Lyapunov functions.

Compressive sensing is a promising solution for the channel estimation in multiple-input multiple-output (MIMO) systems with large antenna arrays and constrained hardware. Utilizing site-specific channel data from real-world systems, deep learning can be employed to learn the compressive sensing measurement vectors with minimum redundancy, thereby focusing sensing power on promising spatial directions of the channel. Collecting real-world channel data, however, is challenging due to the high overhead resulting from the large number of antennas and hardware constraints. In this paper, we propose leveraging a site-specific digital twin to generate synthetic channel data, which shares a similar distribution with real-world data. The synthetic data is then used to train the deep learning models for learning measurement vectors and hybrid precoder/combiner design in an end-to-end manner. We further propose a model refinement approach to fine-tune the model pre-trained on the digital twin data with a small amount of real-world data. The evaluation results show that, by training the model on the digital twin data, the learned measurement vectors can be efficiently adapted to the environment geometry, leading to high performance of hybrid precoding for real-world deployments. Moreover, the model refinement approach can enable the digital twin aided model to achieve comparable performance to the model trained on the real-world dataset with a significantly reduced amount of real-world data., Comment: 7 pages, 5 figures

The correspondence between input text and the generated image exhibits opacity, wherein minor textual modifications can induce substantial deviations in the generated image. While, text embedding, as the pivotal intermediary between text and images, remains relatively underexplored. In this paper, we address this research gap by delving into the text embedding space, unleashing its capacity for controllable image editing and explicable semantic direction attributes within a learning-free framework. Specifically, we identify two critical insights regarding the importance of per-word embedding and their contextual correlations within text embedding, providing instructive principles for learning-free image editing. Additionally, we find that text embedding inherently possesses diverse semantic potentials, and further reveal this property through the lens of singular value decomposition (SVD). These uncovered properties offer practical utility for image editing and semantic discovery. More importantly, we expect the in-depth analyses and findings of the text embedding can enhance the understanding of text-to-image diffusion models.

Supervised image captioning approaches have made great progress, but it is challenging to collect high-quality human-annotated image-text data. Recently, large-scale vision and language models (e.g., CLIP) and large-scale generative language models (e.g., GPT-2) have shown strong performances in various tasks, which also provide some new solutions for image captioning with web paired data, unpaired data or even text-only data. Among them, the mainstream solution is to project image embeddings into the text embedding space with the assistance of consistent representations between image-text pairs from the CLIP model. However, the current methods still face several challenges in adapting to the diversity of data configurations in a unified solution, accurately estimating image-text embedding bias, and correcting unsatisfactory prediction results in the inference stage. This paper proposes a new Text data-centric approach with Interactive Prompts for image Captioning, named TIPCap. 1) We consider four different settings which gradually reduce the dependence on paired data. 2) We construct a mapping module driven by multivariate Gaussian distribution to mitigate the modality gap, which is applicable to the above four different settings. 3) We propose a prompt interaction module that can incorporate optional prompt information before generating captions. Extensive experiments show that our TIPCap outperforms other weakly or unsupervised image captioning methods and achieves a new state-of-the-art performance on two widely used datasets, i.e., MS-COCO and Flickr30K.

Existing hyperspectral image (HSI) super-resolution (SR) methods struggle to effectively capture the complex spectral-spatial relationships and low-level details, while diffusion models represent a promising generative model known for their exceptional performance in modeling complex relations and learning high and low-level visual features. The direct application of diffusion models to HSI SR is hampered by challenges such as difficulties in model convergence and protracted inference time. In this work, we introduce a novel Group-Autoencoder (GAE) framework that synergistically combines with the diffusion model to construct a highly effective HSI SR model (DMGASR). Our proposed GAE framework encodes high-dimensional HSI data into low-dimensional latent space where the diffusion model works, thereby alleviating the difficulty of training the diffusion model while maintaining band correlation and considerably reducing inference time. Experimental results on both natural and remote sensing hyperspectral datasets demonstrate that the proposed method is superior to other state-of-the-art methods both visually and metrically., Comment: Accepted by AAAI2024

Diffusion models have emerged as state-of-the-art generative models for image generation. However, sampling from diffusion models is usually time-consuming due to the inherent autoregressive nature of their sampling process. In this work, we propose a novel approach that accelerates the sampling of diffusion models by parallelizing the autoregressive process. Specifically, we reformulate the sampling process as solving a system of triangular nonlinear equations through fixed-point iteration. With this innovative formulation, we explore several systematic techniques to further reduce the iteration steps required by the solving process. Applying these techniques, we introduce ParaTAA, a universal and training-free parallel sampling algorithm that can leverage extra computational and memory resources to increase the sampling speed. Our experiments demonstrate that ParaTAA can decrease the inference steps required by common sequential sampling algorithms such as DDIM and DDPM by a factor of 4$\sim$14 times. Notably, when applying ParaTAA with 100 steps DDIM for Stable Diffusion, a widely-used text-to-image diffusion model, it can produce the same images as the sequential sampling in only 7 inference steps. The code is available at https://github.com/TZW1998/ParaTAA-Diffusion., Comment: ICML 2024

Reconfigurable intelligent surfaces, with their large number of antennas, offer an interesting opportunity for high spatial-resolution imaging. In this paper, we propose a novel RIS-aided integrated imaging and communication system that can reduce the RIS beam training overhead for communication by leveraging the imaging of the surrounding environment. In particular, using the RIS as a wireless imaging device, our system constructs the scene depth map of the environment, including the mobile user. Then, we develop a user detection algorithm that subtracts the background and extracts the mobile user attributes from the depth map. These attributes are then utilized to design the RIS interaction vector and the beam selection strategy with low overhead. Simulation results show that the proposed approach can achieve comparable beamforming gain to the optimal/exhaustive beam selection solution while requiring 1000 times less beam training overhead., Comment: 6 pages, 4 figures. To appear in Asilomar 2023

In this paper, we explore an integrated sensing and communication (ISAC) system with backscattering RFID tags. In this setup, an access point employs a communication beam to serve a user while leveraging a sensing beam to detect an RFID tag. Under the total transmit power constraint of the system, our objective is to design sensing and communication beams by considering the tag detection and communication requirements. First, we adopt zero-forcing to design the beamforming vectors, followed by solving a convex optimization problem to determine the power allocation between sensing and communication. Then, we study a joint beamforming design problem with the goal of minimizing the total transmit power while satisfying the tag detection and communication requirements. To resolve this, we re-formulate the non-convex constraints into convex second-order cone constraints. The simulation results demonstrate that, under different communication SINR requirements, joint beamforming optimization outperforms the zero-forcing-based method in terms of achievable detection distance, offering a promising approach for the ISAC-backscattering systems., Comment: 5 pages, 5 figures. To appear in IEEE ICC 2024

Dempster-Shafer Theory (DST) as an effective and robust framework for handling uncertain information is applied in decision-making and pattern classification. Unfortunately, its real-time application is limited by the exponential computational complexity. People attempt to address the issue by taking advantage of its mathematical consistency with quantum computing to implement DST operations on quantum circuits and realize speedup. However, the progress so far is still impractical for supporting large-scale DST applications. In this paper, we find that Boolean algebra as an essential mathematical tool bridges the definition of DST and quantum computing. Based on the discovery, we establish a flexible framework mapping any set-theoretically defined DST operations to corresponding quantum circuits for implementation. More critically, this new framework is not only uniform but also enables exponential acceleration for computation and is capable of handling complex applications. Focusing on tasks of classification, we based on a classical attribute fusion algorithm putting forward a quantum evidential classifier, where quantum mass functions for attributes are generated with a simple method and the proposed framework is applied for fusing the attribute evidence. Compared to previous methods, the proposed quantum classifier exponentially reduces the computational complexity to linear. Tests on real datasets validate the feasibility.

Cross-lingual cross-modal retrieval has garnered increasing attention recently, which aims to achieve the alignment between vision and target language (V-T) without using any annotated V-T data pairs. Current methods employ machine translation (MT) to construct pseudo-parallel data pairs, which are then used to learn a multi-lingual and multi-modal embedding space that aligns visual and target-language representations. However, the large heterogeneous gap between vision and text, along with the noise present in target language translations, poses significant challenges in effectively aligning their representations. To address these challenges, we propose a general framework, Cross-Lingual to Cross-Modal (CL2CM), which improves the alignment between vision and target language using cross-lingual transfer. This approach allows us to fully leverage the merits of multi-lingual pre-trained models (e.g., mBERT) and the benefits of the same modality structure, i.e., smaller gap, to provide reliable and comprehensive semantic correspondence (knowledge) for the cross-modal network. We evaluate our proposed approach on two multilingual image-text datasets, Multi30K and MSCOCO, and one video-text dataset, VATEX. The results clearly demonstrate the effectiveness of our proposed method and its high potential for large-scale retrieval., Comment: Accepted by AAAI2024

Coordination engineering was employed to optimize the coordination environment of the Zr atom anchored on the porphyrins (PP). Five promising ZrPP-A candidates as electrocatalysts for nitrogen reduction reaction (NRR) were identified through a four-step screening strategy. First-principles calculations were utilized to evaluate the performance of the candidate electrocatalysts for NRR. A comprehensive search for reaction pathways revealed that NRR reactions with these selected catalysts tend to follow a hybrid pathway. It is found that orbital hybridization and charge transfer between Zr and its coordination atoms, as well as between ZrPP-A and the adsorbed N2 ensured the stability and high catalytic activity of these selected ZrPP-A. Zr plays a crucial role in coordinating charge transfer during the NRR process. Simultaneously, the coordinating atoms and the PP moiety jointly provide additional charge transfers to or from the adsorbate. An asymmetric coordination environment results in an asymmetric charge distribution of the substrate, causing the adsorbed polarized N2 molecule oriented toward the asymmetric charge aggregation region. Our work underscores the importance of considering not only the single-atom catalyst itself but also its coordination environment for the rational design of efficient catalysts.

Text-based Person Search (TBPS) aims to retrieve images of target pedestrian indicated by textual descriptions. It is essential for TBPS to extract fine-grained local features and align them crossing modality. Existing methods utilize external tools or heavy cross-modal interaction to achieve explicit alignment of cross-modal fine-grained features, which is inefficient and time-consuming. In this work, we propose a Vision-Guided Semantic-Group Network (VGSG) for text-based person search to extract well-aligned fine-grained visual and textual features. In the proposed VGSG, we develop a Semantic-Group Textual Learning (SGTL) module and a Vision-guided Knowledge Transfer (VGKT) module to extract textual local features under the guidance of visual local clues. In SGTL, in order to obtain the local textual representation, we group textual features from the channel dimension based on the semantic cues of language expression, which encourages similar semantic patterns to be grouped implicitly without external tools. In VGKT, a vision-guided attention is employed to extract visual-related textual features, which are inherently aligned with visual cues and termed vision-guided textual features. Furthermore, we design a relational knowledge transfer, including a vision-language similarity transfer and a class probability transfer, to adaptively propagate information of the vision-guided textual features to semantic-group textual features. With the help of relational knowledge transfer, VGKT is capable of aligning semantic-group textual features with corresponding visual features without external tools and complex pairwise interaction. Experimental results on two challenging benchmarks demonstrate its superiority over state-of-the-art methods., Comment: Accepted to IEEE TIP

Scanning electron microscopy (SEM) is indispensable in diverse applications ranging from microelectronics to food processing because it provides large depth-of-field images with a resolution beyond the optical diffraction limit. However, the technology requires coating conductive films on insulator samples and a vacuum environment. We use deep learning to obtain the mapping relationship between optical super-resolution (OSR) images and SEM domain images, which enables the transformation of OSR images into SEM-like large depth-of-field images. Our custom-built scanning superlens microscopy (SSUM) system, which requires neither coating samples by conductive films nor a vacuum environment, is used to acquire the OSR images with features down to ~80 nm. The peak signal-to-noise ratio (PSNR) and structural similarity index measure values indicate that the deep learning method performs excellently in image-to-image translation, with a PSNR improvement of about 0.74 dB over the optical super-resolution images. The proposed method provides a high level of detail in the reconstructed results, indicating that it has broad applicability to chip-level defect detection, biological sample analysis, forensics, and various other fields., Comment: 13 pages,7 figures

Dynamic tuning of optical cavities is highly desired in many photonic systems. Here, we show that Fabry-P\'{e}rot(FP) cavities can be actively controlled by the Casimir force. The optical FP cavities consist of a gold nanoplate confronted to an electrical-connecting multi-layer substrate in a liquid environment. The gold nanoplate can be stably suspended due to the balance of repulsive and attractive Casimir forces. Moreover, the suspension distance are modulated strongly by the electric gating or temperature of the system. As a result, we could shift the resonant wavelengthes of the cavities with tens of nanometers at optical frequencies. Finally, we analyze the influence of Brownian motion on the equilibrium distances. Due to the high Q-factor of the FP cavities, our proposed system offers a remarkable platform to experimentally investigate the thermal Casimir effect at sub-micrometer separations, Comment: 6 pages, 5 figures

Pre-trained vision transformers have strong representation benefits to various downstream tasks. Recently, many parameter-efficient fine-tuning (PEFT) methods have been proposed, and their experiments demonstrate that tuning only 1\% extra parameters could surpass full fine-tuning in low-data resource scenarios. However, these methods overlook the task-specific information when fine-tuning diverse downstream tasks. In this paper, we propose a simple yet effective method called "Salient Channel Tuning" (SCT) to leverage the task-specific information by forwarding the model with the task images to select partial channels in a feature map that enables us to tune only 1/8 channels leading to significantly lower parameter costs. Experiments on 19 visual transfer learning downstream tasks demonstrate that our SCT outperforms full fine-tuning on 18 out of 19 tasks by adding only 0.11M parameters of the ViT-B, which is 780$\times$ fewer than its full fine-tuning counterpart. Furthermore, experiments on domain generalization and few-shot classification further demonstrate the effectiveness and generic of our approach. The code is available at https://github.com/showlab/SCT., Comment: This work has been accepted by IJCV

Current research on cross-modal retrieval is mostly English-oriented, as the availability of a large number of English-oriented human-labeled vision-language corpora. In order to break the limit of non-English labeled data, cross-lingual cross-modal retrieval (CCR) has attracted increasing attention. Most CCR methods construct pseudo-parallel vision-language corpora via Machine Translation (MT) to achieve cross-lingual transfer. However, the translated sentences from MT are generally imperfect in describing the corresponding visual contents. Improperly assuming the pseudo-parallel data are correctly correlated will make the networks overfit to the noisy correspondence. Therefore, we propose Dual-view Curricular Optimal Transport (DCOT) to learn with noisy correspondence in CCR. In particular, we quantify the confidence of the sample pair correlation with optimal transport theory from both the cross-lingual and cross-modal views, and design dual-view curriculum learning to dynamically model the transportation costs according to the learning stage of the two views. Extensive experiments are conducted on two multilingual image-text datasets and one video-text dataset, and the results demonstrate the effectiveness and robustness of the proposed method. Besides, our proposed method also shows a good expansibility to cross-lingual image-text baselines and a decent generalization on out-of-domain data.

Unsupervised Contrastive learning has gained prominence in fields such as vision, and biology, leveraging predefined positive/negative samples for representation learning. Data augmentation, categorized into hand-designed and model-based methods, has been identified as a crucial component for enhancing contrastive learning. However, hand-designed methods require human expertise in domain-specific data while sometimes distorting the meaning of the data. In contrast, generative model-based approaches usually require supervised or large-scale external data, which has become a bottleneck constraining model training in many domains. To address the problems presented above, this paper proposes DiffAug, a novel unsupervised contrastive learning technique with diffusion mode-based positive data generation. DiffAug consists of a semantic encoder and a conditional diffusion model; the conditional diffusion model generates new positive samples conditioned on the semantic encoding to serve the training of unsupervised contrast learning. With the help of iterative training of the semantic encoder and diffusion model, DiffAug improves the representation ability in an uninterrupted and unsupervised manner. Experimental evaluations show that DiffAug outperforms hand-designed and SOTA model-based augmentation methods on DNA sequence, visual, and bio-feature datasets. The code for review is released at \url{https://github.com/zangzelin/code_diffaug}., Comment: accepted by ICML24

Person Re-identification (ReID) plays a more and more crucial role in recent years with a wide range of applications. Existing ReID methods are suffering from the challenges of misalignment and occlusions, which degrade the performance dramatically. Most methods tackle such challenges by utilizing external tools to locate body parts or exploiting matching strategies. Nevertheless, the inevitable domain gap between the datasets utilized for external tools and the ReID datasets and the complicated matching process make these methods unreliable and sensitive to noises. In this paper, we propose a Region Generation and Assessment Network (RGANet) to effectively and efficiently detect the human body regions and highlight the important regions. In the proposed RGANet, we first devise a Region Generation Module (RGM) which utilizes the pre-trained CLIP to locate the human body regions using semantic prototypes extracted from text descriptions. Learnable prompt is designed to eliminate domain gap between CLIP datasets and ReID datasets. Then, to measure the importance of each generated region, we introduce a Region Assessment Module (RAM) that assigns confidence scores to different regions and reduces the negative impact of the occlusion regions by lower scores. The RAM consists of a discrimination-aware indicator and an invariance-aware indicator, where the former indicates the capability to distinguish from different identities and the latter represents consistency among the images of the same class of human body regions. Extensive experimental results for six widely-used benchmarks including three tasks (occluded, partial, and holistic) demonstrate the superiority of RGANet against state-of-the-art methods.

Unsupervised domain adaptive person re-identification (Re-ID) methods alleviate the burden of data annotation through generating pseudo supervision messages. However, real-world Re-ID systems, with continuously accumulating data streams, simultaneously demand more robust adaptation and anti-forgetting capabilities. Methods based on image rehearsal addresses the forgetting issue with limited extra storage but carry the risk of privacy leakage. In this work, we propose a Color Prompting (CoP) method for data-free continual unsupervised domain adaptive person Re-ID. Specifically, we employ a light-weighted prompter network to fit the color distribution of the current task together with Re-ID training. Then for the incoming new tasks, the learned color distribution serves as color style transfer guidance to transfer the images into past styles. CoP achieves accurate color style recovery for past tasks with adequate data diversity, leading to superior anti-forgetting effects compared with image rehearsal methods. Moreover, CoP demonstrates strong generalization performance for fast adaptation into new domains, given only a small amount of unlabeled images. Extensive experiments demonstrate that after the continual training pipeline the proposed CoP achieves 6.7% and 8.1% average rank-1 improvements over the replay method on seen and unseen domains, respectively. The source code for this work is publicly available in https://github.com/vimar-gu/ColorPromptReID.

Vision Transformers (ViTs) are normally regarded as a stack of transformer layers. In this work, we propose a novel view of ViTs showing that they can be seen as ensemble networks containing multiple parallel paths with different lengths. Specifically, we equivalently transform the traditional cascade of multi-head self-attention (MSA) and feed-forward network (FFN) into three parallel paths in each transformer layer. Then, we utilize the identity connection in our new transformer form and further transform the ViT into an explicit multi-path ensemble network. From the new perspective, these paths perform two functions: the first is to provide the feature for the classifier directly, and the second is to provide the lower-level feature representation for subsequent longer paths. We investigate the influence of each path for the final prediction and discover that some paths even pull down the performance. Therefore, we propose the path pruning and EnsembleScale skills for improvement, which cut out the underperforming paths and re-weight the ensemble components, respectively, to optimize the path combination and make the short paths focus on providing high-quality representation for subsequent paths. We also demonstrate that our path combination strategies can help ViTs go deeper and act as high-pass filters to filter out partial low-frequency signals. To further enhance the representation of paths served for subsequent paths, self-distillation is applied to transfer knowledge from the long paths to the short paths. This work calls for more future research to explain and design ViTs from new perspectives., Comment: Accepted by ICCV 2023, oral presentation

Utilizing radar sensing for assisting communication has attracted increasing interest thanks to its potential in dynamic environments. A particularly interesting problem for this approach appears in the vehicle-to-vehicle (V2V) millimeter wave and terahertz communication scenarios, where the narrow beams change with the movement of both vehicles. To address this problem, in this work, we develop a radar-aided beam-tracking framework, where a single initial beam and a set of radar measurements over a period of time are utilized to predict the future beams after this time duration. Within this framework, we develop two approaches with the combination of various degrees of radar signal processing and machine learning. To evaluate the feasibility of the solutions in a realistic scenario, we test their performance on a real-world V2V dataset. Our results indicated the importance of high angular resolution radar for this task and affirmed the potential of using radar for the V2V beam management problems., Comment: 5 pages, 5 figures. To appear in EUSIPCO 2023. The dataset is available on the DeepSense 6G website http://deepsense6g.net/