Displays最新文献_第5页

An adaptive U-Net framework for dermatological lesion segmentation 皮肤病变分割的自适应U-Net框架

IF 3.4 2区工程技术

Displays Pub Date : 2026-04-01 Epub Date: 2025-11-17 DOI: 10.1016/j.displa.2025.103290

Ru Huang , Zhimin Qian , Zhengbing Zhou , Zijian Chen , Jiannan Liu , Jing Han , Shuo Zhou , Jianhua He , Xiaoli Chu

{"title":"An adaptive U-Net framework for dermatological lesion segmentation","authors":"Ru Huang , Zhimin Qian , Zhengbing Zhou , Zijian Chen , Jiannan Liu , Jing Han , Shuo Zhou , Jianhua He , Xiaoli Chu","doi":"10.1016/j.displa.2025.103290","DOIUrl":"10.1016/j.displa.2025.103290","url":null,"abstract":"<div><div>With the deep integration of information technology, medical image segmentation has become a crucial tool for dermatological image analysis. However, existing dermatological lesion segmentation methods still face numerous challenges when dealing with complex lesion regions, which result in limited segmentation accuracy. Therefore, this study presents an adaptive segmentation network that draws inspiration from U-Net’s symmetric architecture, with the goal of improving the precision and generalizability of dermatological lesion segmentation. The proposed Visual Scaled Mamba (VSM) module incorporates residual pathways and adaptive scaling factors to enhance fine-grained feature extraction and enable hierarchical representation learning. Additionally, we propose the Multi-Scaled Cross-Axial Attention (MSCA) mechanism, integrating multiscale spatial features and enhancing blurred boundary recognition through dual cross-axial attention. Furthermore, we design an Adaptive Wave-Dilated Bottleneck (AWDB), employing adaptive dilated convolutions and wavelet transforms to improve feature representation and long-range dependency modeling. Through experimental results on the ISIC 2016, ISIC 2018, and PH2 public datasets show that our network achieves a good compromise between model complexity and segmentation accuracy, leading to considerable performance increases in dermatological image segmentation.</div></div>","PeriodicalId":50570,"journal":{"name":"Displays","volume":"92 ","pages":"Article 103290"},"PeriodicalIF":3.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145624546","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Efficient road marking extraction via cooperative enhancement of foundation models and Mamba 通过基础模型和曼巴的协同增强，高效提取道路标线

IF 3.4 2区工程技术

Displays Pub Date : 2026-04-01 Epub Date: 2026-01-14 DOI: 10.1016/j.displa.2026.103351

Kang Zheng, Fu Ren

{"title":"Efficient road marking extraction via cooperative enhancement of foundation models and Mamba","authors":"Kang Zheng, Fu Ren","doi":"10.1016/j.displa.2026.103351","DOIUrl":"10.1016/j.displa.2026.103351","url":null,"abstract":"<div><div>Road marking extraction is critical for high-definition mapping and autonomous driving, yet most lightweight models overlook the long-tailed appearance of thin markings during real-time inference. We propose Efficient Road Markings Segmentation Network (ERMSNet), a hybrid network that pairs lightweight design with the expressive power of Mamba and foundation models. ERMSNet comprises three synergistic branches. (1) A wavelet-augmented Baseline embeds a Road-Marking Mamba (RM-Mamba) whose bi-directional vertical scan captures elongated structures with fewer parameters than vanilla Mamba. (2) A Feature Enhancement branch distills dense image embeddings from the frozen Segment Anything Model (SAM) foundation model through a depth-wise squeeze-and-excitation adapter, injecting rich spatial detail at negligible cost. (3) An Attention Focusing branch projects text–image similarities produced by the Contrastive Language-Image Pre-training (CLIP) foundation model as soft masks that steer the decoder toward rare classes. Comprehensive experiments on CamVid, and our newly released Wuhan Road-Marking (WHRM) benchmark verify the design. Experimental results demonstrate that ERMSNet, with a lightweight configuration of only 0.99 million parameters and 6.44 GFLOPs, achieves mIoU scores of 79.85% and 81.18%, respectively. Compared with existing state-of-the-art methods, ERMSNet significantly reduces computational and memory costs while still delivering outstanding segmentation performance. Its superiority is especially evident in extracting thin and infrequently occurring road marking, highlighting its strong ability to balance efficiency and accuracy. Code and the WHRM dataset will be released upon publication.</div></div>","PeriodicalId":50570,"journal":{"name":"Displays","volume":"92 ","pages":"Article 103351"},"PeriodicalIF":3.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146037277","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Harnessing differentiable geometry and orientation attention for semi-supervised vessel segmentation with limited annotations 利用可微几何和方向注意进行有限注释的半监督血管分割

IF 3.4 2区工程技术

Displays Pub Date : 2026-04-01 Epub Date: 2026-01-21 DOI: 10.1016/j.displa.2026.103347

Yan Liu , Yan Yang , Yongquan Jiang , Xiaole Zhao , Liang Fan

{"title":"Harnessing differentiable geometry and orientation attention for semi-supervised vessel segmentation with limited annotations","authors":"Yan Liu , Yan Yang , Yongquan Jiang , Xiaole Zhao , Liang Fan","doi":"10.1016/j.displa.2026.103347","DOIUrl":"10.1016/j.displa.2026.103347","url":null,"abstract":"<div><div>The precise segmentation of vascular structures is vital for diagnosing retinal and coronary artery diseases. However, the complex morphology and large structural variability of blood vessels make manual annotation time-consuming and finite which in turn limits the scalability of supervised segmentation methods. We propose a semi-supervised segmentation framework named geometric orientational fusion attention network (GOFA-Net) that integrates differentiable geometric augmentation and orientation-aware attention to effectively leverage knowledge from limited annotations. GOFA-Net comprises three key complementary components: 1) a differentiable geometric augmentation strategy (DGAS) employs quaternion-based representations to diversify training samples while preserving prediction consistency between teacher and student models; 2) a multi-view fusion module (MVFM) orchestrates collaborative feature learning between quaternion and conventional convolutional streams to capture comprehensive spatial dependencies; and 3) a global orientational attention module (GOAM) enhances structural awareness through direction-sensitive geometric embeddings, specifically reinforcing the perception of vascular topology along horizontal and vertical orientations. Extensive validation on multiple retinal vessel datasets (DRIVE, STARE, CHASE_DB1, and HRF) and coronary angiography datasets (DCA1 and CHUAC) show that GOFA-Net consistently outperforms state-of-the-art semi-supervised methods, achieving particularly notable gains in scenarios with limited annotations.</div></div>","PeriodicalId":50570,"journal":{"name":"Displays","volume":"92 ","pages":"Article 103347"},"PeriodicalIF":3.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146037274","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

PGgraf: Pose-Guided generative radiance field for novel-views on X-ray PGgraf：姿态引导生成辐射场在x射线上的新观点

IF 3.4 2区工程技术

Displays Pub Date : 2026-04-01 Epub Date: 2026-01-17 DOI: 10.1016/j.displa.2026.103354

Hangyu Li , Moquan Liu , Nan Wang, Mengcheng Sun, Yu Zhu

{"title":"PGgraf: Pose-Guided generative radiance field for novel-views on X-ray","authors":"Hangyu Li , Moquan Liu , Nan Wang, Mengcheng Sun, Yu Zhu","doi":"10.1016/j.displa.2026.103354","DOIUrl":"10.1016/j.displa.2026.103354","url":null,"abstract":"<div><div>In clinical diagnosis, doctors usually judge the information by a few X-rays to avoid excessive ionizing radiation from harming the patient. The recent Neural Radiance Field (NERF) technology contemplates generating novel-views from a single X-ray to assist physicians in diagnosis. In this task, we consider two advantages of X-ray filming over natural images: (1) The medical equipment is fixed, and there is a standardized filming pose. (2) There is an apparent structural prior to X-rays of the same body part at the same pose. Based on such conditions, we propose a Pose-Guided generative radiance field (PGgraf) containing a generator and discriminator. In the training phase, the discriminator combines the image features with two kinds of pose information (ray direction set and camera angle) to guide the generator to synthesize X-rays consistent with the realistic view. In the generator, we design a Density Reconstruction Block (DRB). Unlike the original NERF, which directly estimates the particle density based on the particle positions, the DRB considers all the particle features sampled in a ray and integrally predicts the density of each particle. Experiments comparing qualitative–quantitative on two chest datasets and one knee dataset with state-of-the-art NERF schemes show that PGgraf has a clear advantage in inferring novel-views at different ranges. In the three ranges of 0°to 360°, −15°to 15°, and 75°to 105°, the Peak Signal-to-Noise Ratio (PSNR) improved by an average of 4.18 decibel, and the Learned Perceptual Image Patch Similarity (LPIPS) improved by an average of 50.7%.</div></div>","PeriodicalId":50570,"journal":{"name":"Displays","volume":"92 ","pages":"Article 103354"},"PeriodicalIF":3.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146037272","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Enhanced white efficiency using film color filter via internal reflectance control by capping and refractive index matching layers for rigid OLED panels 采用薄膜滤色片，通过覆盖层和折射率匹配层控制内部反射率，提高了硬OLED面板的白色效率

IF 3.4 2区工程技术

Displays Pub Date : 2026-04-01 Epub Date: 2026-01-09 DOI: 10.1016/j.displa.2026.103345

Horyun Chung , Eunjae Na , Myunghwan Kim , Sungguk An , Yeong Hwan Ko , Jae Su Yu

{"title":"Enhanced white efficiency using film color filter via internal reflectance control by capping and refractive index matching layers for rigid OLED panels","authors":"Horyun Chung , Eunjae Na , Myunghwan Kim , Sungguk An , Yeong Hwan Ko , Jae Su Yu","doi":"10.1016/j.displa.2026.103345","DOIUrl":"10.1016/j.displa.2026.103345","url":null,"abstract":"<div><div>To enhance external luminous efficiency and reduce power consumption in rigid top-emitting organic light-emitting diode (OLED) panels for mobile applications, a film color filter was introduced as a promising alternative for conventional polarizers. The film color filter exhibited higher transmittance in the red, green, and blue emission wavelength regions of OLEDs compared to the polarizer, thereby improving external luminous efficiency. However, its application also increases reflectance due to external light, which necessitates optimization strategies to mitigate this drawback. To address this issue, the internal reflection within the OLED panel was reduced by optimizing the capping layer (CPL) thickness from 60 to 40 nm. Additionally, a refractive index matching layer was implemented between the encapsulation glass and the CPL, resulting in a 24.5% reduction in the specular component included (SCI) reflectance and a decrease in the absolute value of the specular component excluded (SCE) reflection color coordinate. White efficiency typically decreases with the reduction of the CPL thickness; however, the Device B exhibited improvements of 13.7%, 16.8%, and 12.4% in white efficiency compared to the polarizer at the CPL thicknesses of 40, 50, and 60 nm, respectively. This enhancement was particularly pronounced in the blue emission region, where the luminous efficiency is inherently lower. These findings indicate that optimizing the CPL thickness to 40 nm in conjunction with the Device B effectively reduces SCI reflectance, improves SCE reflection color coordinate, and enhances white efficiency. This study demonstrates that replacing the conventional polarizer with a film color filter is a viable approach to achieving higher luminous efficiency in rigid top-emitting OLED panels for mobile devices.</div></div>","PeriodicalId":50570,"journal":{"name":"Displays","volume":"92 ","pages":"Article 103345"},"PeriodicalIF":3.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145976205","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

RES: Reconstruction-based sampling for point cloud learning 基于重构的点云学习采样

IF 3.4 2区工程技术

Displays Pub Date : 2026-04-01 Epub Date: 2025-12-11 DOI: 10.1016/j.displa.2025.103322

Guoqing Zhang, Wenbo Zhao, Junjun Jiang, Xianming Liu

{"title":"RES: Reconstruction-based sampling for point cloud learning","authors":"Guoqing Zhang, Wenbo Zhao, Junjun Jiang, Xianming Liu","doi":"10.1016/j.displa.2025.103322","DOIUrl":"10.1016/j.displa.2025.103322","url":null,"abstract":"<div><div>Processing large-scale point clouds presents a significant challenge. Recent works address this issue by downsampling the point cloud to reduce its size before further processing. However, they often focus on the overall structure of the point cloud, which leads to the loss of small-scale details. Moreover, they typically rely on generic point cloud encoders without considering the characteristics of the downsampled point clouds. In this paper, we re-examine the design of both downsampling and encoder modules and propose a novel reconstruction-based sampling framework for point cloud learning. Specifically, our sampling strategy consists of two components: <em>Point Reconstruction</em> and <em>Shape Reconstruction</em>, which remove points and then reconstruct them from the remaining ones at different detail levels. We then compute the difference between the reconstructed and original points to measure the salience of each point and achieve feature-preserved downsampling by removing points with lower salience. Finally, to efficiently extract features from point clouds, we propose a <em>Local-Global Feature Aggregation</em> (LGFA) module, which first extracts small-scale details through local attention and then captures the overall structure through global attention. Experiments demonstrate that our method achieves outstanding results across various point cloud analysis and downsampling tasks.</div></div>","PeriodicalId":50570,"journal":{"name":"Displays","volume":"92 ","pages":"Article 103322"},"PeriodicalIF":3.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145796851","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

SRINet: Saliency region interaction network for no-reference image quality assessment SRINet：用于无参考图像质量评估的显著性区域交互网络

IF 3.4 2区工程技术

Displays Pub Date : 2026-04-01 Epub Date: 2025-12-12 DOI: 10.1016/j.displa.2025.103317

Maoda Yang, Qicheng Li, Muhan Guo, Yuening Ren, Jun Zhang, Hongxia Deng

{"title":"SRINet: Saliency region interaction network for no-reference image quality assessment","authors":"Maoda Yang, Qicheng Li, Muhan Guo, Yuening Ren, Jun Zhang, Hongxia Deng","doi":"10.1016/j.displa.2025.103317","DOIUrl":"10.1016/j.displa.2025.103317","url":null,"abstract":"<div><div>Image Quality Assessment (IQA) is a fundamental task in computer vision, where existing methods often achieve superior performance by combining global and local representations. Inspired by the mechanism of human visual perception, where focus is placed on visually salient regions when assessing image quality, some studies have attempted to incorporate saliency information as a local feature to assist in quality prediction. However, these methods generally overlook the potential interaction between salient and background regions. To address this issue, we propose a novel IQA method, the Saliency Region Interaction Network (SRINet), which includes saliency-guided feature separation and encoding, region interaction enhancement, and multi-branch fusion. Specifically, image features are first partitioned into salient and background regions using a saliency mask, and each is embedded separately. A regional multi-head attention mechanism is then designed to model the interactive dependencies between these regions. Finally, a cross-attention mechanism, guided by the salient interaction features, fuses this local interactive information with global features, forming a comprehensive, quality-aware model. Experimental results on seven IQA databases demonstrate the competitiveness of SRINet on both synthetically and authentically distorted images.</div></div>","PeriodicalId":50570,"journal":{"name":"Displays","volume":"92 ","pages":"Article 103317"},"PeriodicalIF":3.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145796853","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

The effect of video game history components on virtual and motion sickness susceptibility 视频游戏历史成分对虚拟和晕动病易感性的影响

IF 3.4 2区工程技术

Displays Pub Date : 2026-04-01 Epub Date: 2025-12-20 DOI: 10.1016/j.displa.2025.103327

Stephen J. Fieffer , Jonathan W. Kelly , Michael C. Dorneich , Stephen B. Gilbert

{"title":"The effect of video game history components on virtual and motion sickness susceptibility","authors":"Stephen J. Fieffer , Jonathan W. Kelly , Michael C. Dorneich , Stephen B. Gilbert","doi":"10.1016/j.displa.2025.103327","DOIUrl":"10.1016/j.displa.2025.103327","url":null,"abstract":"<div><div>Visually induced motion sickness (VIMS) and motion sickness (MS) are frequent issues that occur when using screens or moving. While several individual characteristics influence one’s susceptibility to VIMS and MS, one of the more commonly cited factors is video game experience. However, gaming experience is typically considered as a binary variable and its impact has not been systematically characterized. This study employed a large sample (<em>n</em> = 1216) to examine the relationships between components of gaming experience and sickness susceptibility. The Gaming Activity and Motion-sickness Experience and Susceptibility Survey (GAMESS) was created to assess an individual’s status as current and previous gamer, age of first gameplay, and characteristics of typical game play: duration, graphical game type, and game genre. GAMESS also includes brief demographics of age, gender, race, and education, and the short VIMS susceptibility questionnaire (VIMSSQ) and MS susceptibility questionnaire (MSSQ). Current and previous gamers reported significantly lower sickness susceptibility than those who were never gamers, and gamers who played certain game genres (especially first-person shooter and battle royale) reported lower sickness susceptibility than those who played other genres. However, gender differences in sickness susceptibility suggests that gender is a critical factor in the relationship between gaming history and sickness susceptibility. These results highlight the importance of specific details of an individual’s gaming habits and history, going beyond the standard binary classification of gamers and non-gamers to include graphical styles and genres. This research may enable designers of simulated environments to better anticipate the sickness experienced by their users.</div></div>","PeriodicalId":50570,"journal":{"name":"Displays","volume":"92 ","pages":"Article 103327"},"PeriodicalIF":3.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145839194","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

An optimized convolutional neural network based on multi-strategy grey wolf optimizer to identify crop diseases and pests 基于多策略灰狼优化器的优化卷积神经网络作物病虫害识别

IF 3.4 2区工程技术

Displays Pub Date : 2026-04-01 Epub Date: 2026-01-06 DOI: 10.1016/j.displa.2026.103341

Xiaobing Yu , Hongqian Zhang , Yuchen Duan , Xuming Wang

{"title":"An optimized convolutional neural network based on multi-strategy grey wolf optimizer to identify crop diseases and pests","authors":"Xiaobing Yu , Hongqian Zhang , Yuchen Duan , Xuming Wang","doi":"10.1016/j.displa.2026.103341","DOIUrl":"10.1016/j.displa.2026.103341","url":null,"abstract":"<div><div>Agriculture plays a crucial role in national food security, with crop diseases and pests being major threats to agricultural sustainability. Traditional detection methods are labor-intensive, subjective, and often inaccurate. Recent advancements in deep learning have significantly improved image-based recognition; however, the performance of convolutional neural networks (CNNs) is highly dependent on hyperparameter tuning, which remains a challenging <span><span>task. To</span><svg><path></path></svg></span> address this issue, this study proposes a multi-strategy grey wolf optimizer (MGWO) to enhance CNN hyperparameter optimization. MGWO improves the global search efficiency of the conventional grey wolf optimizer (GWO), enabling automatic selection of optimal hyperparameters. The proposed approach is evaluated on corn disease and Pentatomidae stinkbug pest classification, comparing its performance against a baseline CNN model and six other optimization algorithms. Experimental results show that MGWO achieves 95.71% accuracy on the corn disease dataset and 94.46% on the pest dataset, outperforming all competing methods.</div><div>These findings demonstrate the potential of MGWO in optimizing deep learning models for agricultural applications, providing a robust and automated solution for crop disease and pest recognition.</div></div>","PeriodicalId":50570,"journal":{"name":"Displays","volume":"92 ","pages":"Article 103341"},"PeriodicalIF":3.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145938459","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Parameter-efficient fine-tuning for no-reference image quality assessment: Empirical studies on vision transformer 无参考图像质量评估的参数高效微调：视觉变压器的实证研究

IF 3.4 2区工程技术

Displays Pub Date : 2026-04-01 Epub Date: 2026-01-05 DOI: 10.1016/j.displa.2026.103339

GuangLu Sun, Kaiwei Lei, Tianlin Li, Linsen Yu, Suxia Zhu

{"title":"Parameter-efficient fine-tuning for no-reference image quality assessment: Empirical studies on vision transformer","authors":"GuangLu Sun, Kaiwei Lei, Tianlin Li, Linsen Yu, Suxia Zhu","doi":"10.1016/j.displa.2026.103339","DOIUrl":"10.1016/j.displa.2026.103339","url":null,"abstract":"<div><div>Parameter-Efficient Fine-Tuning (PEFT) is a transfer learning technique designed to adapt pre-trained models to downstream tasks while minimizing parameter and computational complexity. In recent years, No-Reference Image Quality Assessment (NR-IQA) methods based on pre-trained visual models have achieved significant progress. However, most of these methods rely on full fine-tuning, which requires substantial computational and memory resources. A natural question arises: can PEFT techniques achieve parameter-efficient NR-IQA with good performance? To explore this, we perform empirical studies using several PEFT methods on pre-trained Vision Transformer (ViT) model. Specifically, we select three PEFT approaches – adapter tuning, prompt tuning, and partial tuning – that have proven effective in general vision tasks, and investigate whether they can achieve performance comparable to traditional visual NR-IQA models. Among them, which is the most effective? Furthermore, we examine the impact of four key factors on the results: fine-tuning position, parameter configuration, layer selection strategy, and the scale of pre-trained weights. Finally, we evaluate whether the optimal PEFT strategy on ViT can be generalized to other Transformer-based architectures. This work offers valuable insights and practical guidance for future research on PEFT methods in NR-IQA tasks.</div></div>","PeriodicalId":50570,"journal":{"name":"Displays","volume":"92 ","pages":"Article 103339"},"PeriodicalIF":3.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145938456","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0