Zhepeng Wang

Jianyu Wen, Jun Xie, Feng Chen et al.

In this paper, we present Self-DACE++, an improved unsupervised and lightweight framework for Low-Light Image Enhancement (LLIE), building upon our previous Self-Reference Deep Adaptive Curve Estimation (Self-DACE). To better address the trade-off between computational efficiency and restoration quality, Self-DACE++ introduces enhanced Adaptive Adjustment Curves (AACs). These curves, governed by minimal trainable parameters, flexibly adjust the dynamic range while preserving the color fidelity, structural integrity, and naturalness of the enhanced images. To achieve an extremely lightweight architecture without sacrificing performance, we propose a randomized order training strategy coupled with a network fusion mechanism, which compresses the model into an efficient iterative inference structure. Furthermore, we formulate a physics-grounded objective function based on Retinex theory and incorporate a dedicated denoising module to effectively estimate and suppress latent noise in dark regions. Extensive qualitative and quantitative evaluations on multiple real-world benchmark datasets demonstrate that Self-DACE++ outperforms existing state-of-the-art methods, delivering superior enhancement quality with real-time inference capability. The code is available at https://github.com/John-Wendell/Self-DACE.

Yawen Wu, Dewen Zeng, Zhepeng Wang et al.

Supervised deep learning needs a large amount of labeled data to achieve high performance. However, in medical imaging analysis, each site may only have a limited amount of data and labels, which makes learning ineffective. Federated learning (FL) can learn a shared model from decentralized data. But traditional FL requires fully-labeled data for training, which is very expensive to obtain. Self-supervised contrastive learning (CL) can learn from unlabeled data for pre-training, followed by fine-tuning with limited annotations. However, when adopting CL in FL, the limited data diversity on each site makes federated contrastive learning (FCL) ineffective. In this work, we propose two federated self-supervised learning frameworks for volumetric medical image segmentation with limited annotations. The first one features high accuracy and fits high-performance servers with high-speed connections. The second one features lower communication costs, suitable for mobile devices. In the first framework, features are exchanged during FCL to provide diverse contrastive data to each site for effective local CL while keeping raw data private. Global structural matching aligns local and remote features for a unified feature space among different sites. In the second framework, to reduce the communication cost for feature exchanging, we propose an optimized method FCLOpt that does not rely on negative samples. To reduce the communications of model download, we propose the predictive target network update (PTNU) that predicts the parameters of the target network. Based on PTNU, we propose the distance prediction (DP) to remove most of the uploads of the target network. Experiments on a cardiac MRI dataset show the proposed two frameworks substantially improve the segmentation and generalization performance compared with state-of-the-art techniques.

Yawen Wu, Dewen Zeng, Zhepeng Wang et al.

Supervised deep learning needs a large amount of labeled data to achieve high performance. However, in medical imaging analysis, each site may only have a limited amount of data and labels, which makes learning ineffective. Federated learning (FL) can help in this regard by learning a shared model while keeping training data local for privacy. Traditional FL requires fully-labeled data for training, which is inconvenient or sometimes infeasible to obtain due to high labeling cost and the requirement of expertise. Contrastive learning (CL), as a self-supervised learning approach, can effectively learn from unlabeled data to pre-train a neural network encoder, followed by fine-tuning for downstream tasks with limited annotations. However, when adopting CL in FL, the limited data diversity on each client makes federated contrastive learning (FCL) ineffective. In this work, we propose an FCL framework for volumetric medical image segmentation with limited annotations. More specifically, we exchange the features in the FCL pre-training process such that diverse contrastive data are provided to each site for effective local CL while keeping raw data private. Based on the exchanged features, global structural matching further leverages the structural similarity to align local features to the remote ones such that a unified feature space can be learned among different sites. Experiments on a cardiac MRI dataset show the proposed framework substantially improves the segmentation performance compared with state-of-the-art techniques.

Kaer Huang, Bingchuan Sun, Feng Chen et al.

In recent years, dominant Multi-object tracking (MOT) and segmentation (MOTS) methods mainly follow the tracking-by-detection paradigm. Transformer-based end-to-end (E2E) solutions bring some ideas to MOT and MOTS, but they cannot achieve a new state-of-the-art (SOTA) performance in major MOT and MOTS benchmarks. Detection and association are two main modules of the tracking-by-detection paradigm. Association techniques mainly depend on the combination of motion and appearance information. As deep learning has been recently developed, the performance of the detection and appearance model is rapidly improved. These trends made us consider whether we can achieve SOTA based on only high-performance detection and appearance model. Our paper mainly focuses on exploring this direction based on CBNetV2 with Swin-B as a detection model and MoCo-v2 as a self-supervised appearance model. Motion information and IoU mapping were removed during the association. Our method wins 1st place on the MOTS track and wins 2nd on the MOT track in the CVPR2023 WAD workshop. We hope our simple and effective method can give some insights to the MOT and MOTS research community. Source code will be released under this git repository

Zhirui Hu, Peiyan Dong, Zhepeng Wang et al.

Model compression, such as pruning and quantization, has been widely applied to optimize neural networks on resource-limited classical devices. Recently, there are growing interest in variational quantum circuits (VQC), that is, a type of neural network on quantum computers (a.k.a., quantum neural networks). It is well known that the near-term quantum devices have high noise and limited resources (i.e., quantum bits, qubits); yet, how to compress quantum neural networks has not been thoroughly studied. One might think it is straightforward to apply the classical compression techniques to quantum scenarios. However, this paper reveals that there exist differences between the compression of quantum and classical neural networks. Based on our observations, we claim that the compilation/traspilation has to be involved in the compression process. On top of this, we propose the very first systematical framework, namely CompVQC, to compress quantum neural networks (QNNs).In CompVQC, the key component is a novel compression algorithm, which is based on the alternating direction method of multipliers (ADMM) approach. Experiments demonstrate the advantage of the CompVQC, reducing the circuit depth (almost over 2.5 %) with a negligible accuracy drop (<1%), which outperforms other competitors. Another promising truth is our CompVQC can indeed promote the robustness of the QNN on the near-term noisy quantum devices.

Benjamin Kiefer, Lojze Žust, Matej Kristan et al.

The 2nd Workshop on Maritime Computer Vision (MaCVi) 2024 addresses maritime computer vision for Unmanned Aerial Vehicles (UAV) and Unmanned Surface Vehicles (USV). Three challenges categories are considered: (i) UAV-based Maritime Object Tracking with Re-identification, (ii) USV-based Maritime Obstacle Segmentation and Detection, (iii) USV-based Maritime Boat Tracking. The USV-based Maritime Obstacle Segmentation and Detection features three sub-challenges, including a new embedded challenge addressing efficicent inference on real-world embedded devices. This report offers a comprehensive overview of the findings from the challenges. We provide both statistical and qualitative analyses, evaluating trends from over 195 submissions. All datasets, evaluation code, and the leaderboard are available to the public at https://macvi.org/workshop/macvi24.

Yawen Wu, Dewen Zeng, Zhepeng Wang et al.

In dermatological disease diagnosis, the private data collected by mobile dermatology assistants exist on distributed mobile devices of patients. Federated learning (FL) can use decentralized data to train models while keeping data local. Existing FL methods assume all the data have labels. However, medical data often comes without full labels due to high labeling costs. Self-supervised learning (SSL) methods, contrastive learning (CL) and masked autoencoders (MAE), can leverage the unlabeled data to pre-train models, followed by fine-tuning with limited labels. However, combining SSL and FL has unique challenges. For example, CL requires diverse data but each device only has limited data. For MAE, while Vision Transformer (ViT) based MAE has higher accuracy over CNNs in centralized learning, MAE's performance in FL with unlabeled data has not been investigated. Besides, the ViT synchronization between the server and clients is different from traditional CNNs. Therefore, special synchronization methods need to be designed. In this work, we propose two federated self-supervised learning frameworks for dermatological disease diagnosis with limited labels. The first one features lower computation costs, suitable for mobile devices. The second one features high accuracy and fits high-performance servers. Based on CL, we proposed federated contrastive learning with feature sharing (FedCLF). Features are shared for diverse contrastive information without sharing raw data for privacy. Based on MAE, we proposed FedMAE. Knowledge split separates the global and local knowledge learned from each client. Only global knowledge is aggregated for higher generalization performance. Experiments on dermatological disease datasets show superior accuracy of the proposed frameworks over state-of-the-arts.

Jinyang Li, Zhepeng Wang, Zhirui Hu et al.

Quantum computing presents a promising approach for machine learning with its capability for extremely parallel computation in high-dimension through superposition and entanglement. Despite its potential, existing quantum learning algorithms, such as Variational Quantum Circuits(VQCs), face challenges in handling more complex datasets, particularly those that are not linearly separable. What's more, it encounters the deployability issue, making the learning models suffer a drastic accuracy drop after deploying them to the actual quantum devices. To overcome these limitations, this paper proposes a novel spatial-temporal design, namely ST-VQC, to integrate non-linearity in quantum learning and improve the robustness of the learning model to noise. Specifically, ST-VQC can extract spatial features via a novel block-based encoding quantum sub-circuit coupled with a layer-wise computation quantum sub-circuit to enable temporal-wise deep learning. Additionally, a SWAP-Free physical circuit design is devised to improve robustness. These designs bring a number of hyperparameters. After a systematic analysis of the design space for each design component, an automated optimization framework is proposed to generate the ST-VQC quantum circuit. The proposed ST-VQC has been evaluated on two IBM quantum processors, ibm_cairo with 27 qubits and ibmq_lima with 7 qubits to assess its effectiveness. The results of the evaluation on the standard dataset for binary classification show that ST-VQC can achieve over 30% accuracy improvement compared with existing VQCs on actual quantum computers. Moreover, on a non-linear synthetic dataset, the ST-VQC outperforms a linear classifier by 27.9%, while the linear classifier using classical computing outperforms the existing VQC by 15.58%.

Zhepeng Wang, Jinyang Li, Zhirui Hu et al.

Security has always been a critical issue in machine learning (ML) applications. Due to the high cost of model training -- such as collecting relevant samples, labeling data, and consuming computing power -- model-stealing attack is one of the most fundamental but vitally important issues. When it comes to quantum computing, such a quantum machine learning (QML) model-stealing attack also exists and is even more severe because the traditional encryption method, such as homomorphic encryption can hardly be directly applied to quantum computation. On the other hand, due to the limited quantum computing resources, the monetary cost of training QML model can be even higher than classical ones in the near term. Therefore, a well-tuned QML model developed by a third-party company can be delegated to a quantum cloud provider as a service to be used by ordinary users. In this case, the QML model will likely be leaked if the cloud provider is under attack. To address such a problem, we propose a novel framework, namely QuMoS, to preserve model security. We propose to divide the complete QML model into multiple parts and distribute them to multiple physically isolated quantum cloud providers for execution. As such, even if the adversary in a single provider can obtain a partial model, it does not have sufficient information to retrieve the complete model. Although promising, we observed that an arbitrary model design under distributed settings cannot provide model security. We further developed a reinforcement learning-based security engine, which can automatically optimize the model design under the distributed setting, such that a good trade-off between model performance and security can be made. Experimental results on four datasets show that the model design proposed by QuMoS can achieve competitive performance while providing the highest security than the baselines.

Zhepeng Wang, Runxue Bao, Yawen Wu et al.

Pretrained large language models (LLMs) have revolutionized natural language processing (NLP) tasks such as summarization, question answering, and translation. However, LLMs pose significant security risks due to their tendency to memorize training data, leading to potential privacy breaches and copyright infringement. Accurate measurement of this memorization is essential to evaluate and mitigate these potential risks. However, previous attempts to characterize memorization are constrained by either using prefixes only or by prepending a constant soft prompt to the prefixes, which cannot react to changes in input. To address this challenge, we propose a novel method for estimating LLM memorization using dynamic, prefix-dependent soft prompts. Our approach involves training a transformer-based generator to produce soft prompts that adapt to changes in input, thereby enabling more accurate extraction of memorized data. Our method not only addresses the limitations of previous methods but also demonstrates superior performance in diverse experimental settings compared to state-of-the-art techniques. In particular, our method can achieve the maximum relative improvement of 112.75% and 32.26% over the vanilla baseline in terms of discoverable memorization rate for the text generation task and code generation task respectively.

Yang Zhou, Can Jin, Zihan Dong et al.

Reinforcement learning improves the reasoning ability of large language models but remains costly and sample-inefficient, as many rollouts provide weak learning signals. Difficulty-aware data selection methods attempt to address this by prioritizing moderately difficult prompts, yet our analysis reveals three limitations: difficulty estimates become inaccurate under policy drift, data selection alone yields limited final-performance gains, and inference efficiency remains largely unchanged. These findings suggest that efficient and effective RL requires more than filtering by difficulty: the policy should learn to solve hard tasks while producing concise responses for easy ones. To this end, we propose **Dare**, a unified framework that co-evolves difficulty estimation with the policy via self-normalized importance sampling, maintains diverse difficulty coverage through a symmetric Beta sampling distribution, and applies tailored training strategies across difficulty tiers with adaptive compute allocation. Extensive experiments across multiple models and domains demonstrate that **Dare** consistently outperforms existing methods in training efficiency, final effectiveness, and inference efficiency, producing more concise responses on easy tasks while improving correctness on hard ones. Code is available at https://github.com/EtaYang10th/DARE.

Hongzhu Yi, Xinming Wang, Zhenghao zhang et al.

Within the domain of large language models, reinforcement fine-tuning algorithms necessitate the generation of a complete reasoning trajectory beginning from the input query, which incurs significant computational overhead during the rollout phase of training. To address this issue, we analyze the impact of different segments of the reasoning path on the correctness of the final result and, based on these insights, propose Reinforcement Fine-Tuning with Partial Reasoning Optimization (RPO), a plug-and-play reinforcement fine-tuning algorithm. Unlike traditional reinforcement fine-tuning algorithms that generate full reasoning paths, RPO trains the model by generating suffixes of the reasoning path using experience cache. During the rollout phase of training, RPO reduces token generation in this phase by approximately 95%, greatly lowering the theoretical time overhead. Compared with full-path reinforcement fine-tuning algorithms, RPO reduces the training time of the 1.5B model by 90% and the 7B model by 72%. At the same time, it can be integrated with typical algorithms such as GRPO and DAPO, enabling them to achieve training acceleration while maintaining performance comparable to the original algorithms. Our code is open-sourced at https://github.com/yhz5613813/RPO.

Zhepeng Wang, Feng Chen, Kanokphan Lertniphonphan et al.

This report presents our Le3DE2E solution for unified sensor-based detection, tracking, and forecasting in Argoverse Challenges at CVPR 2023 Workshop on Autonomous Driving (WAD). We propose a unified network that incorporates three tasks, including detection, tracking, and forecasting. This solution adopts a strong Bird's Eye View (BEV) encoder with spatial and temporal fusion and generates unified representations for multi-tasks. The solution was tested in the Argoverse 2 sensor dataset to evaluate the detection, tracking, and forecasting of 26 object categories. We achieved 1st place in Detection, Tracking, and Forecasting on the E2E Forecasting track in Argoverse Challenges at CVPR 2023 WAD.

Zhepeng Wang, Isaacshubhanand Putla, Weiwen Jiang et al.

Seismic full waveform inversion (FWI) is a widely used technique in geophysics for inferring subsurface structures from seismic data. And InversionNet is one of the most successful data-driven machine learning models that is applied to seismic FWI. However, the high computing costs to run InversionNet have made it challenging to be efficiently deployed on edge devices that are usually resource-constrained. Therefore, we propose to employ the structured pruning algorithm to get a lightweight version of InversionNet, which can make an efficient inference on edge devices. And we also made a prototype with Raspberry Pi to run the lightweight InversionNet. Experimental results show that the pruned InversionNet can achieve up to 98.2 % reduction in computing resources with moderate model performance degradation.

Liang Tang, Shuxian Li, Yuhao Cheng et al.

Computer use agent is an emerging area in artificial intelligence that aims to operate the computers to achieve the user's tasks, which attracts a lot of attention from both industry and academia. However, the present agents' performance is far from being used. In this paper, we propose the Self-Evolution Agent (SEA) for computer use, and to develop this agent, we propose creative methods in data generation, reinforcement learning, and model enhancement. Specifically, we first propose an automatic pipeline to generate the verifiable trajectory for training. And then, we propose efficient step-wise reinforcement learning to alleviate the significant computational requirements for long-horizon training. In the end, we propose the enhancement method to merge the grounding and planning ability into one model without any extra training. Accordingly, based on our proposed innovation of data generation, training strategy, and enhancement, we get the Selfevolution Agent (SEA) for computer use with only 7B parameters, which outperforms models with the same number of parameters and has comparable performance to larger ones. We will make the models' weight and related codes open-source in the future.

Shijing Wang, Yaping Huang, Jun Xie et al.

Achieving accurate and reliable gaze predictions in complex and diverse environments remains challenging. Fortunately, it is straightforward to access diverse gaze datasets in real-world applications. We discover that training these datasets jointly can significantly improve the generalization of gaze estimation, which is overlooked in previous works. However, due to the inherent distribution shift across different datasets, simply mixing multiple dataset decreases the performance in the original domain despite gaining better generalization abilities. To address the problem of ``cross-dataset gaze estimation'', we propose a novel Evidential Inter-intra Fusion EIF framework, for training a cross-dataset model that performs well across all source and unseen domains. Specifically, we build independent single-dataset branches for various datasets where the data space is partitioned into overlapping subspaces within each dataset for local regression, and further create a cross-dataset branch to integrate the generalizable features from single-dataset branches. Furthermore, evidential regressors based on the Normal and Inverse-Gamma (NIG) distribution are designed to additionally provide uncertainty estimation apart from predicting gaze. Building upon this foundation, our proposed framework achieves both intra-evidential fusion among multiple local regressors within each dataset and inter-evidential fusion among multiple branches by Mixture \textbfof Normal Inverse-Gamma (MoNIG distribution. Experiments demonstrate that our method consistently achieves notable improvements in both source domains and unseen domains.

Jun Xie, Yingjian Zhu, Feng Chen et al.

In this paper, we present our solution for the semi-supervised learning track (MER-SEMI) in MER2025. We propose a comprehensive framework, grounded in the principle that "more is better," to construct a robust Mixture of Experts (MoE) emotion recognition system. Our approach integrates a diverse range of input modalities as independent experts, including novel signals such as knowledge from large Vision-Language Models (VLMs) and temporal Action Unit (AU) information. To effectively utilize unlabeled data, we introduce a consensus-based pseudo-labeling strategy, generating high-quality labels from the agreement between a baseline model and Gemini, which are then used in a two-stage training paradigm. Finally, we employ a multi-expert voting ensemble combined with a rule-based re-ranking process to correct prediction bias and better align the outputs with human preferences. Evaluated on the MER2025-SEMI challenge dataset, our method achieves an F1-score of 0.8772 on the test set, ranking 2nd in the track. Our code is available at https://github.com/zhuyjan/MER2025-MRAC25.

Jun Xie, Xiongjun Guan, Yingjian Zhu et al.

In this paper, we present the runner-up solution for the Ego4D EgoSchema Challenge at CVPR 2025 (Confirmed on May 20, 2025). Inspired by the success of large models, we evaluate and leverage leading accessible multimodal large models and adapt them to video understanding tasks via few-shot learning and model ensemble strategies. Specifically, diversified prompt styles and process paradigms are systematically explored and evaluated to effectively guide the attention of large models, fully unleashing their powerful generalization and adaptability abilities. Experimental results demonstrate that, with our carefully designed approach, directly utilizing an individual multimodal model already outperforms the previous state-of-the-art (SOTA) method which includes several additional processes. Besides, an additional stage is further introduced that facilitates the cooperation and ensemble of periodic results, which achieves impressive performance improvements. We hope this work serves as a valuable reference for the practical application of large models and inspires future research in the field. Our Code is available at https://github.com/XiongjunGuan/EgoSchema-CVPR25.

Zhenghao Zhang, Jun Xie, Xingchen Chen et al.

The prevalence of real-world multi-view data makes incomplete multi-view clustering (IMVC) a crucial research. The rapid development of Graph Neural Networks (GNNs) has established them as one of the mainstream approaches for multi-view clustering. Despite significant progress in GNNs-based IMVC, some challenges remain: (1) Most methods rely on the K-Nearest Neighbors (KNN) algorithm to construct static graphs from raw data, which introduces noise and diminishes the robustness of the graph topology. (2) Existing methods typically utilize the Mean Squared Error (MSE) loss between the reconstructed graph and the sparse adjacency graph directly as the graph reconstruction loss, leading to substantial gradient noise during optimization. To address these issues, we propose a novel \textbf{D}ynamic Deep \textbf{G}raph Learning for \textbf{I}ncomplete \textbf{M}ulti-\textbf{V}iew \textbf{C}lustering with \textbf{M}asked Graph Reconstruction Loss (DGIMVCM). Firstly, we construct a missing-robust global graph from the raw data. A graph convolutional embedding layer is then designed to extract primary features and refined dynamic view-specific graph structures, leveraging the global graph for imputation of missing views. This process is complemented by graph structure contrastive learning, which identifies consistency among view-specific graph structures. Secondly, a graph self-attention encoder is introduced to extract high-level representations based on the imputed primary features and view-specific graphs, and is optimized with a masked graph reconstruction loss to mitigate gradient noise during optimization. Finally, a clustering module is constructed and optimized through a pseudo-label self-supervised training mechanism. Extensive experiments on multiple datasets validate the effectiveness and superiority of DGIMVCM.

Pengfei Zheng, Kanokphan Lertniphonphan, Feng Chen et al.

This report presents our Le3DE2E_Occ solution for 4D Occupancy Forecasting in Argoverse Challenges at CVPR 2023 Workshop on Autonomous Driving (WAD). Our solution consists of a strong LiDAR-based Bird's Eye View (BEV) encoder with temporal fusion and a two-stage decoder, which combines a DETR head and a UNet decoder. The solution was tested on the Argoverse 2 sensor dataset to evaluate the occupancy state 3 seconds in the future. Our solution achieved 18% lower L1 Error (3.57) than the baseline and got the 1 place on the 4D Occupancy Forecasting task in Argoverse Challenges at CVPR 2023.

Kanokphan Lertniphonphan, Feng Chen, Junda Xu et al.

This report presents our team's PCIE_Interaction solution for the Ego4D Social Interaction Challenge at CVPR 2025, addressing both Looking At Me (LAM) and Talking To Me (TTM) tasks. The challenge requires accurate detection of social interactions between subjects and the camera wearer, with LAM relying exclusively on face crop sequences and TTM combining speaker face crops with synchronized audio segments. In the LAM track, we employ face quality enhancement and ensemble methods. For the TTM task, we extend visual interaction analysis by fusing audio and visual cues, weighted by a visual quality score. Our approach achieved 0.81 and 0.71 mean average precision (mAP) on the LAM and TTM challenges leader board. Code is available at https://github.com/KanokphanL/PCIE_Ego4D_Social_Interaction

Feng Chen, Ling Ding, Kanokphan Lertniphonphan et al.

This report presents our team's 'PCIE_EgoHandPose' solution for the EgoExo4D Hand Pose Challenge at CVPR2024. The main goal of the challenge is to accurately estimate hand poses, which involve 21 3D joints, using an RGB egocentric video image provided for the task. This task is particularly challenging due to the subtle movements and occlusions. To handle the complexity of the task, we propose the Hand Pose Vision Transformer (HP-ViT). The HP-ViT comprises a ViT backbone and transformer head to estimate joint positions in 3D, utilizing MPJPE and RLE loss function. Our approach achieved the 1st position in the Hand Pose challenge with 25.51 MPJPE and 8.49 PA-MPJPE. Code is available at https://github.com/KanokphanL/PCIE_EgoHandPose

Kanokphan Lertniphonphan, Jun Xie, Yaqing Meng et al.

This report presents our team's 'PCIE_LAM' solution for the Ego4D Looking At Me Challenge at CVPR2024. The main goal of the challenge is to accurately determine if a person in the scene is looking at the camera wearer, based on a video where the faces of social partners have been localized. Our proposed solution, InternLSTM, consists of an InternVL image encoder and a Bi-LSTM network. The InternVL extracts spatial features, while the Bi-LSTM extracts temporal features. However, this task is highly challenging due to the distance between the person in the scene and the camera movement, which results in significant blurring in the face image. To address the complexity of the task, we implemented a Gaze Smoothing filter to eliminate noise or spikes from the output. Our approach achieved the 1st position in the looking at me challenge with 0.81 mAP and 0.93 accuracy rate. Code is available at https://github.com/KanokphanL/Ego4D_LAM_InternLSTM

Yuliang Liu, Sheng Zhang, Lianwen Jin et al.

Scene text in the wild is commonly presented with high variant characteristics. Using quadrilateral bounding box to localize the text instance is nearly indispensable for detection methods. However, recent researches reveal that introducing quadrilateral bounding box for scene text detection will bring a label confusion issue which is easily overlooked, and this issue may significantly undermine the detection performance. To address this issue, in this paper, we propose a novel method called Sequential-free Box Discretization (SBD) by discretizing the bounding box into key edges (KE) which can further derive more effective methods to improve detection performance. Experiments showed that the proposed method can outperform state-of-the-art methods in many popular scene text benchmarks, including ICDAR 2015, MLT, and MSRA-TD500. Ablation study also showed that simply integrating the SBD into Mask R-CNN framework, the detection performance can be substantially improved. Furthermore, an experiment on the general object dataset HRSC2016 (multi-oriented ships) showed that our method can outperform recent state-of-the-art methods by a large margin, demonstrating its powerful generalization ability. Source code: https://github.com/Yuliang-Liu/Box_Discretization_Network.

Ziying Song, Guoxin Zhang, Jun Xie et al.

LiDAR-camera fusion can enhance the performance of 3D object detection by utilizing complementary information between depth-aware LiDAR points and semantically rich images. Existing voxel-based methods face significant challenges when fusing sparse voxel features with dense image features in a one-to-one manner, resulting in the loss of the advantages of images, including semantic and continuity information, leading to sub-optimal detection performance, especially at long distances. In this paper, we present VoxelNextFusion, a multi-modal 3D object detection framework specifically designed for voxel-based methods, which effectively bridges the gap between sparse point clouds and dense images. In particular, we propose a voxel-based image pipeline that involves projecting point clouds onto images to obtain both pixel- and patch-level features. These features are then fused using a self-attention to obtain a combined representation. Moreover, to address the issue of background features present in patches, we propose a feature importance module that effectively distinguishes between foreground and background features, thus minimizing the impact of the background features. Extensive experiments were conducted on the widely used KITTI and nuScenes 3D object detection benchmarks. Notably, our VoxelNextFusion achieved around +3.20% in AP@0.7 improvement for car detection in hard level compared to the Voxel R-CNN baseline on the KITTI test dataset

Zhepeng Wang, Yi Sheng, Nirajan Koirala et al.

Benefiting from cloud computing, today's early-stage quantum computers can be remotely accessed via the cloud services, known as Quantum-as-a-Service (QaaS). However, it poses a high risk of data leakage in quantum machine learning (QML). To run a QML model with QaaS, users need to locally compile their quantum circuits including the subcircuit of data encoding first and then send the compiled circuit to the QaaS provider for execution. If the QaaS provider is untrustworthy, the subcircuit to encode the raw data can be easily stolen. Therefore, we propose a co-design framework for preserving the data security of QML with the QaaS paradigm, namely PristiQ. By introducing an encryption subcircuit with extra secure qubits associated with a user-defined security key, the security of data can be greatly enhanced. And an automatic search algorithm is proposed to optimize the model to maintain its performance on the encrypted quantum data. Experimental results on simulation and the actual IBM quantum computer both prove the ability of PristiQ to provide high security for the quantum data while maintaining the model performance in QML.

Lei Lu, Zhepeng Wang, Runxue Bao et al.

Existing pruning techniques for large language models (LLMs) targeting domain-specific applications typically follow a two-stage process: pruning the pretrained general-purpose LLMs and then fine-tuning the pruned LLMs on specific domains. However, the pruning decisions, derived from the pretrained weights, remain unchanged during fine-tuning, even if the weights have been updated. Therefore, such a combination of the pruning decisions and the finetuned weights may be suboptimal, leading to non-negligible performance degradation. To address these limitations, we propose ATP: All-in-One Tuning and Structural Pruning, a unified one-stage structural pruning and fine-tuning approach that dynamically identifies the current optimal substructure throughout the fine-tuning phase via a trainable pruning decision generator. Moreover, given the limited available data for domain-specific applications, Low-Rank Adaptation (LoRA) becomes a common technique to fine-tune the LLMs. In ATP, we introduce LoRA-aware forward and sparsity regularization to ensure that the substructures corresponding to the learned pruning decisions can be directly removed after the ATP process. ATP outperforms the state-of-the-art two-stage pruning methods on tasks in the legal and healthcare domains. More specifically, ATP recovers up to 88% and 91% performance of the dense model when pruning 40% parameters of LLaMA2-7B and LLaMA3-8B models, respectively.

Zhepeng Wang, Runxue Bao, Yawen Wu et al.

Graph neural networks (GNNs) are powerful machine learning models designed to handle irregularly structured data. However, their generic design often proves inadequate for analyzing brain connectomes in Alzheimer's Disease (AD), highlighting the need to incorporate domain knowledge for optimal performance. Infusing AD-related knowledge into GNNs is a complicated task. Existing methods typically rely on collaboration between computer scientists and domain experts, which can be both time-intensive and resource-demanding. To address these limitations, this paper presents a novel self-guided, knowledge-infused multimodal GNN that autonomously incorporates domain knowledge into the model development process. Our approach conceptualizes domain knowledge as natural language and introduces a specialized multimodal GNN capable of leveraging this uncurated knowledge to guide the learning process of the GNN, such that it can improve the model performance and strengthen the interpretability of the predictions. To evaluate our framework, we curated a comprehensive dataset of recent peer-reviewed papers on AD and integrated it with multiple real-world AD datasets. Experimental results demonstrate the ability of our method to extract relevant domain knowledge, provide graph-based explanations for AD diagnosis, and improve the overall performance of the GNN. This approach provides a more scalable and efficient alternative to inject domain knowledge for AD compared with the manual design from the domain expert, advancing both prediction accuracy and interpretability in AD diagnosis.

Zhepeng Wang, Yingjian Zhu, Guanghao Dong et al.

This study investigates the integration of trustworthy prior reasoning knowledge from MLLMs into multimodal emotion recognition. We employ Gemini to generate fine-grained, modality-separable reasoning traces, which are injected as priors during the fusion stage to enrich cross-modal interactions. To mitigate the pronounced class-imbalance in multimodal emotion recognition, we introduce Balanced Dual-Contrastive Learning, a loss formulation that jointly balances inter-class and intra-class distributions. Applied to the MER2024 benchmark, our prior-enhanced framework yields substantial performance gains, demonstrating that the reliability of MLLM-derived reasoning can be synergistically combined with the domain adaptability of lightweight fusion networks for robust, scalable emotion recognition.

Jun Xie, Zhaoran Zhao, Xiongjun Guan et al.

We propose a novel framework for open-ended video question answering that enhances reasoning depth and robustness in complex real-world scenarios, as benchmarked on the CVRR-ES dataset. Existing Video-Large Multimodal Models (Video-LMMs) often exhibit limited contextual understanding, weak temporal modeling, and poor generalization to ambiguous or compositional queries. To address these challenges, we introduce a prompting-and-response integration mechanism that coordinates multiple heterogeneous Video-Language Models (VLMs) via structured chains of thought, each tailored to distinct reasoning pathways. An external Large Language Model (LLM) serves as an evaluator and integrator, selecting and fusing the most reliable responses. Extensive experiments demonstrate that our method significantly outperforms existing baselines across all evaluation metrics, showcasing superior generalization and robustness. Our approach offers a lightweight, extensible strategy for advancing multimodal reasoning without requiring model retraining, setting a strong foundation for future Video-LMM development.

Feng Chen, Kanokphan Lertniphonphan, Qiancheng Yan et al.

This report introduces our team's (PCIE_EgoPose) solutions for the EgoExo4D Pose and Proficiency Estimation Challenges at CVPR2025. Focused on the intricate task of estimating 21 3D hand joints from RGB egocentric videos, which are complicated by subtle movements and frequent occlusions, we developed the Hand Pose Vision Transformer (HP-ViT+). This architecture synergizes a Vision Transformer and a CNN backbone, using weighted fusion to refine the hand pose predictions. For the EgoExo4D Body Pose Challenge, we adopted a multimodal spatio-temporal feature integration strategy to address the complexities of body pose estimation across dynamic contexts. Our methods achieved remarkable performance: 8.31 PA-MPJPE in the Hand Pose Challenge and 11.25 MPJPE in the Body Pose Challenge, securing championship titles in both competitions. We extended our pose estimation solutions to the Proficiency Estimation task, applying core technologies such as transformer-based architectures. This extension enabled us to achieve a top-1 accuracy of 0.53, a SOTA result, in the Demonstrator Proficiency Estimation competition.

Yawen Wu, Zhepeng Wang, Dewen Zeng et al.

Contrastive learning (CL), a self-supervised learning approach, can effectively learn visual representations from unlabeled data. Given the CL training data, generative models can be trained to generate synthetic data to supplement the real data. Using both synthetic and real data for CL training has the potential to improve the quality of learned representations. However, synthetic data usually has lower quality than real data, and using synthetic data may not improve CL compared with using real data. To tackle this problem, we propose a data generation framework with two methods to improve CL training by joint sample generation and contrastive learning. The first approach generates hard samples for the main model. The generator is jointly learned with the main model to dynamically customize hard samples based on the training state of the main model. Besides, a pair of data generators are proposed to generate similar but distinct samples as positive pairs. In joint learning, the hardness of a positive pair is progressively increased by decreasing their similarity. Experimental results on multiple datasets show superior accuracy and data efficiency of the proposed data generation methods applied to CL. For example, about 4.0%, 3.5%, and 2.6% accuracy improvements for linear classification are observed on ImageNet-100, CIFAR-100, and CIFAR-10, respectively. Besides, up to 2x data efficiency for linear classification and up to 5x data efficiency for transfer learning are achieved.

Yawen Wu, Dewen Zeng, Zhepeng Wang et al.

Deep learning models have been deployed in an increasing number of edge and mobile devices to provide healthcare. These models rely on training with a tremendous amount of labeled data to achieve high accuracy. However, for medical applications such as dermatological disease diagnosis, the private data collected by mobile dermatology assistants exist on distributed mobile devices of patients, and each device only has a limited amount of data. Directly learning from limited data greatly deteriorates the performance of learned models. Federated learning (FL) can train models by using data distributed on devices while keeping the data local for privacy. Existing works on FL assume all the data have ground-truth labels. However, medical data often comes without any accompanying labels since labeling requires expertise and results in prohibitively high labor costs. The recently developed self-supervised learning approach, contrastive learning (CL), can leverage the unlabeled data to pre-train a model, after which the model is fine-tuned on limited labeled data for dermatological disease diagnosis. However, simply combining CL with FL as federated contrastive learning (FCL) will result in ineffective learning since CL requires diverse data for learning but each device only has limited data. In this work, we propose an on-device FCL framework for dermatological disease diagnosis with limited labels. Features are shared in the FCL pre-training process to provide diverse and accurate contrastive information. After that, the pre-trained model is fine-tuned with local labeled data independently on each device or collaboratively with supervised federated learning on all devices. Experiments on dermatological disease datasets show that the proposed framework effectively improves the recall and precision of dermatological disease diagnosis compared with state-of-the-art methods.

Yawen Wu, Zhepeng Wang, Dewen Zeng et al.

Collaborative learning enables distributed clients to learn a shared model for prediction while keeping the training data local on each client. However, existing collaborative learning methods require fully-labeled data for training, which is inconvenient or sometimes infeasible to obtain due to the high labeling cost and the requirement of expertise. The lack of labels makes collaborative learning impractical in many realistic settings. Self-supervised learning can address this challenge by learning from unlabeled data. Contrastive learning (CL), a self-supervised learning approach, can effectively learn visual representations from unlabeled image data. However, the distributed data collected on clients are usually not independent and identically distributed (non-IID) among clients, and each client may only have few classes of data, which degrades the performance of CL and learned representations. To tackle this problem, we propose a collaborative contrastive learning framework consisting of two approaches: feature fusion and neighborhood matching, by which a unified feature space among clients is learned for better data representations. Feature fusion provides remote features as accurate contrastive information to each client for better local learning. Neighborhood matching further aligns each client's local features to the remote features such that well-clustered features among clients can be learned. Extensive experiments show the effectiveness of the proposed framework. It outperforms other methods by 11% on IID data and matches the performance of centralized learning.

Zhepeng Wang, Zhiding Liang, Shanglin Zhou et al.

With the constant increase of the number of quantum bits (qubits) in the actual quantum computers, implementing and accelerating the prevalent deep learning on quantum computers are becoming possible. Along with this trend, there emerge quantum neural architectures based on different designs of quantum neurons. A fundamental question in quantum deep learning arises: what is the best quantum neural architecture? Inspired by the design of neural architectures for classical computing which typically employs multiple types of neurons, this paper makes the very first attempt to mix quantum neuron designs to build quantum neural architectures. We observe that the existing quantum neuron designs may be quite different but complementary, such as neurons from variational quantum circuits (VQC) and Quantumflow. More specifically, VQC can apply real-valued weights but suffer from being extended to multiple layers, while QuantumFlow can build a multi-layer network efficiently, but is limited to use binary weights. To take their respective advantages, we propose to mix them together and figure out a way to connect them seamlessly without additional costly measurement. We further investigate the design principles to mix quantum neurons, which can provide guidance for quantum neural architecture exploration in the future. Experimental results demonstrate that the identified quantum neural architectures with mixed quantum neurons can achieve 90.62% of accuracy on the MNIST dataset, compared with 52.77% and 69.92% on the VQC and QuantumFlow, respectively.

Zhiding Liang, Zhepeng Wang, Junhuan Yang et al.

In the noisy intermediate-scale quantum (NISQ) era, one of the key questions is how to deal with the high noise level existing in physical quantum bits (qubits). Quantum error correction is promising but requires an extensive number (e.g., over 1,000) of physical qubits to create one "perfect" qubit, exceeding the capacity of the existing quantum computers. This paper aims to tackle the noise issue from another angle: instead of creating perfect qubits for general quantum algorithms, we investigate the potential to mitigate the noise issue for dedicate algorithms. Specifically, this paper targets quantum neural network (QNN), and proposes to learn the errors in the training phase, so that the identified QNN model can be resilient to noise. As a result, the implementation of QNN needs no or a small number of additional physical qubits, which is more realistic for the near-term quantum computers. To achieve this goal, an application-specific compiler is essential: on the one hand, the error cannot be learned if the mapping from logical qubits to physical qubits exists randomness; on the other hand, the compiler needs to be efficient so that the lengthy training procedure can be completed in a reasonable time. In this paper, we utilize the recent QNN framework, QuantumFlow, as a case study. Experimental results show that the proposed approach can optimize QNN models for different errors in qubits, achieving up to 28% accuracy improvement compared with the model obtained by the error-agnostic training.

Yawen Wu, Zhepeng Wang, Dewen Zeng et al.

After a model is deployed on edge devices, it is desirable for these devices to learn from unlabeled data to continuously improve accuracy. Contrastive learning has demonstrated its great potential in learning from unlabeled data. However, the online input data are usually none independent and identically distributed (non-iid) and storages of edge devices are usually too limited to store enough representative data from different data classes. We propose a framework to automatically select the most representative data from the unlabeled input stream, which only requires a small data buffer for dynamic learning. Experiments show that accuracy and learning speed are greatly improved.

Zhenge Jia, Zhepeng Wang, Feng Hong et al.

Life-threatening ventricular arrhythmias (VA) are the leading cause of sudden cardiac death (SCD), which is the most significant cause of natural death in the US. The implantable cardioverter defibrillator (ICD) is a small device implanted to patients under high risk of SCD as a preventive treatment. The ICD continuously monitors the intracardiac rhythm and delivers shock when detecting the life-threatening VA. Traditional methods detect VA by setting criteria on the detected rhythm. However, those methods suffer from a high inappropriate shock rate and require a regular follow-up to optimize criteria parameters for each ICD recipient. To ameliorate the challenges, we propose the personalized computing framework for deep learning based VA detection on medical IoT systems. The system consists of intracardiac and surface rhythm monitors, and the cloud platform for data uploading, diagnosis, and CNN model personalization. We equip the system with real-time inference on both intracardiac and surface rhythm monitors. To improve the detection accuracy, we enable the monitors to detect VA collaboratively by proposing the cooperative inference. We also introduce the CNN personalization for each patient based on the computing framework to tackle the unlabeled and limited rhythm data problem. When compared with the traditional detection algorithm, the proposed method achieves comparable accuracy on VA rhythm detection and 6.6% reduction in inappropriate shock rate, while the average inference latency is kept at 71ms.

Yawen Wu, Zhepeng Wang, Yiyu Shi et al.

This work aims to enable on-device training of convolutional neural networks (CNNs) by reducing the computation cost at training time. CNN models are usually trained on high-performance computers and only the trained models are deployed to edge devices. But the statically trained model cannot adapt dynamically in a real environment and may result in low accuracy for new inputs. On-device training by learning from the real-world data after deployment can greatly improve accuracy. However, the high computation cost makes training prohibitive for resource-constrained devices. To tackle this problem, we explore the computational redundancies in training and reduce the computation cost by two complementary approaches: self-supervised early instance filtering on data level and error map pruning on the algorithm level. The early instance filter selects important instances from the input stream to train the network and drops trivial ones. The error map pruning further prunes out insignificant computations when training with the selected instances. Extensive experiments show that the computation cost is substantially reduced without any or with marginal accuracy loss. For example, when training ResNet-110 on CIFAR-10, we achieve 68% computation saving while preserving full accuracy and 75% computation saving with a marginal accuracy loss of 1.3%. Aggressive computation saving of 96% is achieved with less than 0.1% accuracy loss when quantization is integrated into the proposed approaches. Besides, when training LeNet on MNIST, we save 79% computation while boosting accuracy by 0.2%.

Yawen Wu, Zhepeng Wang, Zhenge Jia et al.

This work aims to enable persistent, event-driven sensing and decision capabilities for energy-harvesting (EH)-powered devices by deploying lightweight DNNs onto EH-powered devices. However, harvested energy is usually weak and unpredictable and even lightweight DNNs take multiple power cycles to finish one inference. To eliminate the indefinite long wait to accumulate energy for one inference and to optimize the accuracy, we developed a power trace-aware and exit-guided network compression algorithm to compress and deploy multi-exit neural networks to EH-powered microcontrollers (MCUs) and select exits during execution according to available energy. The experimental results show superior accuracy and latency compared with state-of-the-art techniques.