政府計畫(GRB),建議「依年度遞減排序」,以查看最新的研究方向。
畢業學年度 | 論文標題 | 連結 | 學位 | 畢業時長(years) |
---|---|---|---|---|
關鍵字 | ||||
113 | 基於視覺語... 基於視覺語言模型知識蒸餾與三元特徵精煉之無源域領域自適應 (Vision-Language Distillation and Triplet Feature Refinement for Source-Free Domain Adaptation) | NTHU NDLTD | 碩 | 2.04 |
領域自適應(Source-Free Domain Adaptation)、視覺語言模型(Vision-Language model)、領域不變(Domain invariant)、偽標籤(Pseudo-labeling)、知識蒸餾(Knowledge distillation)、三重特徵精煉(Triplet feature refinement) 領域自適應... | ||||
113 | 基於提示的... 基於提示的多模態對比學習於跨壓縮率的深度偽造檢測 (Prompt-guided Multi-modal contrastive learning for Cross-compression-rate Deepfake Detection) | NTHU NDLTD | 碩 | 2.04 |
深度偽造檢測(Deepfake Detection)、多模態(Multi-modal)、對比學習(Contrastive learning)、跨壓縮率(Cross-compression-rate) 深度偽造檢... | ||||
112 | 基於多模態... 基於多模態轉換引導特徵學習的多模態人臉防偽技術 (Generalized Multimodal Face Anti-Spoofing via Multimodal Transition-Guided Feature Learning) | NTHU NDLTD | 碩(外籍生) | 2.00 |
人臉反欺騙(Face Anti-Spoofing)、模式轉變(Modalities Transition)、多模式(Multi Modalities)、單類學習(One class) 人臉反欺騙... | ||||
112 | 通過合成信... 通過合成信號引導特徵學習的全測試時間 rPPG 預測 (Fully Test-Time rPPG Estimation via Synthetic Signal-Guided Feature Learning) | NTHU NDLTD | 碩 | 2.00 |
遠程光電容積脈搏波(rPPG)、遠距離心律偵測(Remote Heart Rate estimation)、測試時間適應(Test-time adaptation)、深度學習(Deep learning) 遠程光電容... | ||||
112 | SLIP:... SLIP: 透過語言引導的活體特徵學習進行單一類別人臉防偽 (SLIP: One-Class Face Anti-Spoofing via Language-Guided Liveness Feature Learning) | NTHU NDLTD | 碩 | 2.00 |
人臉防偽(face anti-spoofing)、電腦視覺(computer vision)、單類別(one-class)、語言引導(language-guided) 人臉防偽(... | ||||
112 | 自監督式特... 自監督式特徵分離學習與擴增於單領域泛化的遠程心率估計 (Self-Supervised Disentangled Feature Learning and Augmentation for Single Domain Generalization in rPPG Estimation) | NTHU NDLTD | 碩 | 2.18 |
遠距光體積變化描記圖(Remote photoplethysmography)、單領域泛化(Single-domain generalization)、自監督特徵分離(Self-supervised feature disentanglement)、資料擴增(Data augmentation) 遠距光體積... | ||||
111 | 朝向多樣化... 朝向多樣化的活體特徵表示和跨域人臉反偽造的域擴展 (Towards Diverse Liveness Feature Representation and Domain Expansion for Cross-Domain Face Anti-Spoofing) | NTHU NDLTD | 碩 | 1.99 |
人臉防偽造(face anti-spoofing)、域泛化(domain generalization)、特徵解耦學習(disentangled feature learning)、仿射特徵轉換(affine feature transformation)、對抗學習(adversarial learning) 人臉防偽造... | ||||
111 | 測試時間適... 測試時間適應於強健的人臉反偽造 (Test-Time Adaptation for Robust Face Anti-Spoofing) | NTHU NDLTD | 碩 | 1.99 |
測試時間適應(Test-Time Adaptation)、人臉反偽造(Face Anti-Spoofing) 測試時間適... | ||||
111 | 基於雙重域... 基於雙重域擴增進行單一域泛化語意分割 (Single-Domain Generalization for Semantic Segmentation via Dual-Level Domain Augmentation) | NTHU NDLTD | 碩 | 1.99 |
單一域泛化(single-domain generalization)、語意分割(semantic segmentation) 單一域泛化... | ||||
111 | 通過使用三... 通過使用三元組挖掘以及補丁增強的人臉反欺騙域泛化 (Domain Generalization for Face Anti-spoofing via Patch-wise Augmentation and Triplet Mining) | NTHU NDLTD | 碩 | 2.15 |
人臉防偽(face anti-spoofing)、域泛化(domain generalization)、補丁增強(patch-wise augmentation)、三源組挖掘(triplet mining) 人臉防偽(... | ||||
111 | 提升穩健領... 提升穩健領域內與領域外泛化能力: 利用原型對齊與協作注意力模組之對比式學習 (Towards Robust In-Domain and Out-of-Domain Generalization: Contrastive Learning with Prototype Alignment and Collaborative Attention) | NTHU NDLTD | 碩 | 2.01 |
領域泛化(Domain generalization)、對比式學習(Contrastive learning)、嘈雜標籤(Noisy labels)、度量學習(Metric learning) 領域泛化(... | ||||
111 | 基於分佈轉... 基於分佈轉移與適應性類平衡自學習的無源域適應語意分割 (Source-Free Domain Adaptation for Semantic Segmentation via Distribution Transfer and Adaptive Class-Balanced Self-Training) | NTHU NDLTD | 碩 | 2.01 |
語意分割(Semantic Segmentation)、無源域適應(Source-Free Domain Adaptation)、域適應(Domain Adaptation)、負學習(Negative Learning)、類平衡(Class Imbalance) 語意分割(... | ||||
110 | 領域泛化遠... 領域泛化遠程心率估計網路:具有領域置換和領域擴增的分離特徵學習 (Domain Generalized RPPG Network: Disentangled Feature Learning with Domain Permutation and Domain Augmentation) | NTHU NDLTD | 碩(提早入學) | 2.48 |
遠程心率估計(rPPG estimation)、領域泛化(domain generalization)、領域擴增(domain augmentation)、解離式學習(disentanglement) 遠程心率估... | ||||
110 | rPPG基... rPPG基準數據集的增強:透過雙循環一致學習從未成對的臉部影像學習消除和嵌入rPPG 信號 (Augmentation of rPPG Benchmark Datasets: Learning to Remove and Embed rPPG Signals via Double Cycle Consistent Learning from Unpaired Facial Videos) | NTHU NDLTD | 碩 | 1.98 |
遠程光體積變化描記圖(Remote Photoplethysmography)、資料增強(Data Augmentation)、雙循環一致學習(Double-Cycle Consistency)、遠程心率偵測(Remote Heart Rate Measurement) 遠程光體積... | ||||
109 | 基於強烈輔... 基於強烈輔助預測以及有區別的特徵學習進行人臉反偽造偵測 (Face Anti-Spoofing via Robust Auxiliary Estimation and Discriminative Feature Learning) | NTHU NDLTD | 碩 | 2.15 |
人臉反偽造(Face anti-spoofing)、對抗學習(Adversarial learning)、對比損失(Contrastive loss) 人臉反偽造... | ||||
109 | TFT 面... TFT 面板瑕疵分類之半監督式域自適應學習實證研究 (Semi-Supervised Domain Adaptation for LCD TFT Defect Classification) | NTHU NDLTD | 碩 | 2.13 |
面板瑕疵分類(Semi-Supervised)、半監督式域自適應學習(Domain) 面板瑕疵分... | ||||
109 | 用於領域泛... 用於領域泛化之具有輔助語義學習的特徵空間中之風格替換 (Style Replacement in Feature Space with Auxiliary Semantic Learning for Domain Generalization) | NTHU NDLTD | 碩 | 2.09 |
領域泛化(Domain generalization)、風格替換(Style replacement)、風格轉換(Style transfer)、自監督學習(Self-supervised learning) 領域泛化(... | ||||
109 | 趨向目標:... 趨向目標:自我正規化漸進式學習於語義分割之無監督領域適應 (Towards the Target: Self-Regularized Progressive Learning for Unsupervised Domain Adaptation on Semantic Segmentation) | NTHU NDLTD | 碩 | 2.04 |
無監督領域適應(Unsupervised Domain Adaptation)、語義分割(Semantic Segmentation)、漸進式學習(Progressive Learning)、標註放寬(Label Relaxation)、類別不平衡(Class Imbalance) 無監督領域... | ||||
109 | 領域自適應... 領域自適應語義分割的自引導對抗學習 (Self-guided Adversarial Learning for Domain Adaptive Semantic Segmentation) | NTHU NDLTD | 碩 | 2.04 |
非監督式之領域自適應(Unsupervised domain adaptation)、語意分割(Semantic segmentation)、自引導對抗式學習(Self-guided adversarial learning) 非監督式之... | ||||
109 | 基於廣泛風... 基於廣泛風格與關注特徵進行域泛化的風格擴充方法 (Style Augmentation for Domain Generalization using Diverse Styles and Attention Content Features) | NTHU NDLTD | 碩 | 2.12 |
風格擴充(Style Augmentation)、風格轉換(Style Transfer)、資料擴充(Data Augmentation)、域泛化(Domain Generalization)、關注模塊(Attention Module)、廣泛風格(Diverse Styles) 風格擴充(... | ||||
109 | 空間注意力... 空間注意力機制同步化影片生成和遠程光體積變化描記圖法 (Simultaneous Video Generation and Remote Photoplethysmography Estimation with Spatial Attention Mechanism) | NTHU NDLTD | 碩 | 2.12 |
影片生成(Video generation)、遠程光體積變化描記圖法(Remote photoplethysmography)、計算機視覺(Computer vision)、深度學習(Deep learning)、人工智慧(Artificial intelligence)、心率偵測(Heart rate estimation)、生醫影像(Biomedical imaging)、資料強化(Data augmentation)、注意力機制(Attention mechanism) 影片生成(... | ||||
109 | 同步生成影... 同步生成影片和估計遠程光體積變化描計圖的多任務學習 (Multi-Task Learning for Simultaneous Video Generation and Remote Photoplethysmography Estimation) | NTHU NDLTD | 碩 | 2.11 |
計算遠程光體積變化描記圖(Remote photoplethysmography estimation)、影片生成(Video generation)、多任務學習(Multi-task learning) 計算遠程光... | ||||
109 | 學習去概括... 學習去概括: 用於領域均化的類感知對抗式學習 (Learning to Generalize: Class-Aware Adversarial Learning for Domain Generalization) | NTHU NDLTD | 碩 | 2.11 |
對抗式學習(adversarial learning)、類感知(class-aware)、元學習(meta-learning)、領域泛化(domain generalization) 對抗式學習... | ||||
107 | 應用於遠端... 應用於遠端心率偵測的雙通道時間網路 (Siamese Temporal Network for Remote Heart Rate Estimation) | NTHU NDLTD | 碩 | 1.96 |
遠端心率偵測(Remote heart rate estimation)、孿生網絡(Siamese network)、卷積類神經網路(Convolutional neural network) 遠端心率偵... | ||||
107 | 利用遠程光... 利用遠程光體積變化描記圖之特徵達成基於視覺之心率偵測 (Exploiting Remote Photoplethysmography Features for Vision-based Heart Rate Estimation) | NTHU NDLTD | 碩 | 1.96 |
遠程心率偵測(Remote heart rate estimation)、低秩(Low-rank)、遠程光體積變化描記圖(Remote photoplethysmography) 遠程心率偵... | ||||
107 | 應用於影片... 應用於影片中動作辨識之運動和外觀的分解表示法學習 (Decomposed Representation Learning of Motion and Appearance for Video Action Recognition) | NTHU NDLTD | 碩 | 1.96 |
動作辨識(Action Recognition)、分解表示法學習(Decomposed Representation Learning)、運動表示法(Motion Representation) 動作辨識(... | ||||
107 | 應用於場景... 應用於場景分類之語意感知的關注模組 (SAAM: Semantic-Aware Attention Module for Scene Classification) | NTHU NDLTD | 碩 | 1.96 |
場景分類(Scene Classification)、關注模組(Attention Module)、共同關注(Co-attention) 場景分類(... | ||||
107 | 應用於連續... 應用於連續情緒識別之時空特徵學習 (Disentangled and Spatiotemporal Feature Learning for Continuous Emotion Recognition) | NTHU NDLTD | 碩 | 2.05 |
連續情感識別(Continuous Emotion Recognition)、時空特徵學習(Convolutional Neural Network)、對抗式網路學習(Disentangled feature learning GAN)、捲積神經網路(Long Short Term Memory)、長短期記憶(Emotion dimension) 連續情感識... | ||||
107 | 應用於小樣... 應用於小樣本影像辨識之輸入適應之度量學習與具區別性特徵選取 (Query-Adapted Metric Learning and Discriminative Feature Selection for Few-Shot Recognition) | NTHU NDLTD | 碩 | 2.05 |
深度學習(Few-shot recognition)、小樣本影像辨識(Deep learning)、度量學習(Metric learning)、特徵選取(Few-shot learning)、小樣本學習(Feature selection) 深度學習(... | ||||
106 | 應用於域自... 應用於域自適應學習之生成對抗引導網絡 (Generative Adversarial Guided Learning for Domain Adaptation) | NTHU NDLTD | 碩 | 1.98 |
深度學習(Deep learning)、電腦視覺(Computer vision)、域自適應學習(Domain adaptation)、生成對抗網絡(Generative adversarial network) 深度學習(... | ||||
106 | 基於生成自... 基於生成自我引導的密集標籤之弱監督語義分割 (Generating Self-Guided Dense Annotations for Weakly Supervised Semantic Segmentation) | NTHU NDLTD | 碩 | 1.98 |
弱監督(Weakly Supervised)、語義分割(Semantic Segmentation)、自我引導(Self-Guided) 弱監督(W... | ||||
105 | 以標籤共同... 以標籤共同偏移估計與標籤信賴先驗之連續表情識別研究 (Continuous Emotion Recognition by Estimating Common Label Bias with Label Confidence Prior) | NTHU NDLTD | 碩 | 暫無口試日期 |
連續情緒識別(Continuous Emotion Recognition)、機器學習(Machine Learning)、人臉表情識別(Facial Expression Recognition) 連續情緒識... | ||||
105 | 基於音框層... 基於音框層級自動標註預測之繪圖式音樂檢索系統 (Sketch-based Music Retrieval Based on Frame-level Auto-tagging Predictions) | NTHU NDLTD | 碩 | 1.98 |
音樂檢索(music retrieval)、繪圖式檢索(sketch-based retrieval)、人機互動(human-computer interaction)、自動標註(auto-tagging) 音樂檢索(... | ||||
105 | 應用於即時... 應用於即時情感辨識之基於屬性的雙通道時間網絡 (AST-Net: An Attribute-based Siamese Temporal Network for Real-Time Emotion Recognition) | NTHU NDLTD | 碩 | 1.98 |
情感辨識(Emotion Recognition)、時間網絡(Temporal Network)、卷積類神經網路(Attribute Feature) 情感辨識(... | ||||
105 | 應用於連續... 應用於連續情緒辨識之基於殘差影像的情緒特徵學習 (Emotion-Related Feature Learning From Residual Images for Continuous Emotion Recognition) | NTHU NDLTD | 碩 | 1.98 |
連續情緒辨識(Continuous)、特徵學習(Emotion)、殘差影像(Recognition)、捲積神經網路(Residual)、長短期記憶(Image) 連續情緒辨... | ||||
105 | 類別共享卷... 類別共享卷積網路: 基於像素級實例邊界從語意分割切割物件 (Class-Shared Convolutional Neural Network for Instance Segmentation via Pixelwise Instance Boundary) | NTHU NDLTD | 碩 | 1.96 |
實例語意分割(Instance)、像素級實例邊界(Semantic)、卷積網路(Segmentation) 實例語意分... | ||||
104 | 藉由深度學... 藉由深度學習預測商品類別再結合細部外觀以及局部組態之商品辨識 (Mobile Product Recognition with Deep Learning of Category Selection, Fine-grained Appearance and Part Configuration) | NTHU NDLTD | 碩 | 暫無口試日期 |
商品辨識(Mobile product recognition)、卷積類神經網路(Convolutional neural network)、聚類(Clustering)、細部外觀描述(Fine-grained appearance) 商品辨識(... | ||||
104 | 透過跨層併... 透過跨層併列與多尺度預測的完全卷積網路之語意分割 (Fully Convolutional Networks with Cross-layer Concatenation and Multi-Scale Prediction for Semantic Segmentation) | NTHU NDLTD | 碩 | 暫無口試日期 |
語意分割(Semantic segmentation)、深度學習(Deep learning)、卷積網絡(Convolutional neural network) 語意分割(... | ||||
104 | 結合監督式... 結合監督式學習的知識及可轉移資訊之真實世界場景分析 (Semantic Segmentation for Real-World Data by Jointly Exploiting Supervised and Transferrable Knowledge) | NTHU NDLTD | 碩 | 暫無口試日期 |
場景分析(semantic segmentation)、真實世界資料庫(real-world dataset)、標籤轉移(label transfer) 場景分析(... | ||||
104 | 透過深度類... 透過深度類神經網路結合細部外觀變化以及局部組態達到商品辨識之目的 (Mobile Product Recognition by Involving Fine-Grained Appearance and Part Configuration into Deep Neural Network) | NTHU NDLTD | 碩 | 暫無口試日期 |
商品辨識(production recognition)、深度學習(deep learning) 商品辨識(... | ||||
103 | 利用深度卷... 利用深度卷積特徵及密集對齊之非參數場景分析 (Nonparametric Scene Parsing with Deep Convolutional Features and Dense Alignment) | NTHU NDLTD | 碩 | 暫無口試日期 |
場景分析(scene parsing)、物體視窗(object window)、深度卷積式網路(deep convolutional network)、SIFT flow(SIFT flow) 場景分析(... | ||||
103 | 透過人臉區... 透過人臉區域校正與階梯式深度網路的人臉特徵點偵測 (Facial Landmark Detection with Face Region Rectification and Deep Network Cascade) | NTHU NDLTD | 碩 | 暫無口試日期 |
人臉特徵點偵測(facial landmark detection)、深度學習(deep learning)、卷積類神經網路(convolutional neural network)、階梯式(cascade) 人臉特徵點... | ||||
102 | 基於姿勢感... 基於姿勢感知之深度卷積網路的人臉特徵點偵測 (Facial Landmark Detection using Pose-Aware Deep Convolutional Network) | NTHU NDLTD | 碩 | 暫無口試日期 |
人臉特徵點偵測、深度學習、卷積神經網路 人臉特徵點... | ||||
102 | 基於低秩稀... 基於低秩稀疏分解進行特徵混合之局部影像竄改偵測 (A Feature Fusion Model with Rank-Sparsity Decomposition for Image Tampering Localization) | NTHU NDLTD | 碩 | 暫無口試日期 |
局部影像竄改偵測、特徵混合、低秩稀疏分解 局部影像竄... | ||||
101 | 雙子空間非... 雙子空間非負圖形嵌入表示法及其於人臉表情辨識的應用 (Dual Subspace Nonnegative Graph Embedding for Person-Invariant Facial Expression Recognition) | NTHU NDLTD | 碩 | 暫無口試日期 |
非負矩陣分解(Nonnegative Matrix Factorization)、臉部表情辨識(Facial Expression Recognition)、圖形嵌入表示法(Graph-Embedding)、雙子空間(Dual Subspace) 非負矩陣分... | ||||
101 | 不需對齊之... 不需對齊之成對影像曝光融合 (Exposure Fusion of Image Pair without Alignment) | NTHU NDLTD | 碩 | 暫無口試日期 |
影像曝光融合(Exposure Fusion)、曝光融合(Alignment)、曝光合成(HDR)、高動態(undefined)、對齊(undefined) 影像曝光融... | ||||
101 | 透過稀疏與... 透過稀疏與低張量秩建模之移動物件偵測 (Moving Object Detection via Sparse and Low-Rank Tensor Modeling) | NTHU NDLTD | 碩 | 暫無口試日期 |
移動物件偵測(moving object detection)、低張量秩(low-rank tensor) 移動物件偵... | ||||
101 | 運用結構性... 運用結構性區域限制修正最佳化光線傳播率圖之單張影像去霧 (Single Image Dehazing Using Optimal Transmission Under Textured-Region Constraint) | NTHU NDLTD | 碩 | 暫無口試日期 |
去霧(Dehazing)、光線傳播率(Transmission)、結構性區域限制(Textured-region constraint) 去霧(De... |