帥宏翰

Enhancing Text-to-Image Diffusion Models Under Special Prompts via Training-Free Inference-Time Optimization

電機工程學系

文字生成圖像、擴散模型、免訓練推論、罕見概念提示詞、長提示詞多樣性、分數替換、高斯混合取樣、text-to-image generation、diffusion models、training-free inference、rare-concept prompts、long-prompt diversity、score replacement、mixture-of-Gaussians sampling

InCoP: A Benchmark for Indoor Collaborative 3D Object Detection with Complementary Ground-Robot Views

電機工程學系

協作感知、三維物件偵測、室內機器人、多模態融合、互補視角、通訊效率、collaborative perception、3D object detection、indoor robotics、multimodal fusion、complementary views、communication efficiency

FlatFlow TTS: Single-NFE Speech Synthesis via Consistency Distillation and EPSS-Enhanced MeanFlow

人工智慧技術與應用碩士學位學程

流匹配、一致性蒸餾、MeanFlow、單步語音合成、知識蒸餾、EPSS、flow matching、consistency distillation、single-step speech synthesis、knowledge distillation

Wardrobe Wizardry：Versatile Dataset Generation for High-Resolution Virtual Try-On

電機工程學系

虛擬試穿、深度學習、虛擬試穿資料集、影像合成、Virtual Try-On、Deep Learning、Virtual Try-On Dataset、Image Synthesis

Multimodal Large Language Model Guided Diffusion for Visual Synthesis and Editing

電機學院博士班・共同指導：鄭文皇

擴散模型、多模態大語言模型、多模態學習、影像生成、影像編輯、Diffusion Models、Multimodal Large Language Models、Multimodal Learning、Image Generation、Image Editing

On Gradient Norm Constraints and Preference-Based Learning in Generative Models

電機工程學系

生成對抗網路、梯度穩定性、擴散模型、逆向強化學習、偏好學習、Generative adversarial networks、Gradient stability、Diffusion models、Inverse reinforcement learning、Preference alignment

Seeing Beyond Time: Vision-Language Intelligence Inspired by Human Perception of Future, Present, and Past

電機工程學系

人工智慧、深度學習、時序特徵學習、軌跡預測、場景理解、視覺語言模型、Artificial Intelligence、Deep Learning、Temporal Representation Learning、Trajectory Forecasting、Scene Understanding、Vision-Language Models

When Anchors Meet Cold Diffusion: A Multi-Stage Approach to Lane Detection

電機工程學系

深度學習、電腦視覺、車道偵測、擴散模型、Deep Learning、Computer Vision、Lane Detection、Diffusion Model

Learnable Query-Enhanced Pose Transformation

智能系統研究所

擴散模型、姿態引導的圖像生成、可學習查詢、深度學習、電腦視覺、Diffusion Model、Pose-Guided Image Synthesis、Learnable Queries、Deep Learning、Computer Vision

RADIANCE：Relative Adaptive Denoising with IP-Adapter for Novel Concept Enhancement

電機工程學系

圖片生成、組合式生成、回饋式引導、無需訓練、擴散模型、Image Generation、Compositional Generation、Feedback-based Guidance、Trainingfree、Diffusion Model

A Study of Deep Learning across Challenging Visual Domains

電機學院博士班

深度學習、領域適應、合成孔徑雷達目標識別、知識蒸餾、遮罩自編碼器、小樣本學習、基礎模型、deep learning、domain adaptation、SAR target recognition、knowledge distillation、masked autoencoder (MAE)、few-shot learning、foundation model

To Memorize or Not to Memorize: Backdoor-Driven Data Leakage in Diffusion Models

電機工程學系

擴散模型、後門攻擊、資料外洩、模型安全性、生成式模型、Diffusion Model、Backdoor Attack、Data Leakage、Model Security、Generative Model

Adversarial Attacks on VQA-NLE: Exposing and Alleviating Inconsistencies in Visual Question Answering Explanations

資訊安全研究所

對抗式攻擊、穩健性、模型偏差、視覺問答、解釋忠實度、自然語言解釋、Adversarial Attacks、Robustness、Model Bias、Vision Question Answering、Explanation Faithfulness、Natural Language Explanations

Blur2Sharp: Human Novel Pose and View Synthesis with Generative Prior Refinement

電機工程學系・共同指導：黃敬群

新視角生成、新姿勢生成、神經渲染、擴散模型、單張影像人體生成、Novel Pose Synthesis、Novel View Synthesis、Neural Rendering、Diffusion Models、Single-Image Human Generation

IterDiff: Training-Free Iterative Face Editing via Efficient CLIP-guided Memory Bank

電機工程學系

深度學習、電腦視覺、擴散模型、圖像編輯、Deep Learning、Computer Vision、Diffusion Model、Image Editing

Evolving Frameworks for Object Detection: From Tiny Targets to 3D Scenes

電機工程學系・共同指導：鄭文皇

小物件辨識、單目三維物件辨識、Tiny Object Detection、Monocular 3D Object Detection

FPW: Frequency-domain Pixel-by-Pixel Watermarking against unauthorized images used on training generative model

電機工程學系

浮水印、隱蔽性、深度學習、機械學習、生成模型、擴散模型、Watermark、Invisibility、Deep Learning、Machine Learning、Generative Model、Diffusion Model

From Diffusion to Decision: A Diffusion-ReRanking in Scene Text Detection

電機工程學系

場景文字、檢測方法、擴散模型、重排序、實例分割、Scene Text、Detection、Diffusion-based、Reranking、Instance Segmentation

Coarse-to-fine Refinement for Fine-grained Visual Recognition

電機工程學系・共同指導：鄭文皇

細粒度視覺分類、視覺情感識別、多模態大語言模型、細粒度分佈外檢測、Fine-grained Visual Classification、Visual Emotion Recognition、Multi-modal Large Language Models、Fine-grained Out-of-distribution Detection

Improving Camouflage Object Segmentation via Multi-Scale Global-to-Local Feature Integration

電機資訊國際博士學位學程

偽裝物件分割、偽裝實例分割、全域到局部互動、Camouflage Object Segmentation、Camouflage Instance Segmentation、Global-to-Local Interaction

Enhancing Coherence and Fidelity in Camouflaged Image Generation via Foreground Focus

人工智慧技術與應用碩士學位學程・共同指導：鄭文皇

偽裝圖像生成、潛在擴散模型、前景聚焦、和諧性增強、保真度維持、Camouflaged Image Generation、Latent Diffusion Model、Foreground Focus、Coherence Enhancement、Fidelity Preservation

A Universal Framework for Heterogeneous Knowledge Distillation based on Region-Aware Attention and Adaptive Feedback Prompt

智能系統研究所・共同指導：鄭文皇

電腦視覺、知識蒸餾、異質架構、提示微調、視覺轉換器、多層感知機、卷積神經網路、Computer vision、Knowledge distillation、Heterogeneous architecture、Prompt tuning、Vision transformer、Multi-layer perceptron layer、Convolution neural network

Embracing Your Scale: Scale-Aware Offset Feature Alignment for Crowd Flux Estimation

人工智慧技術與應用碩士學位學程

深度學習、電腦視覺、人群流量估測、人群計數、多尺度架構、Deep learning、computer vision、crowd flux estimation、crowd counting、multi-scale architecture

DetailSemNet: Elevating Signature Verification with Captured Details and Semantics by Feature Disentanglement and Re-entanglement

人工智慧技術與應用碩士學位學程・共同指導：黃敬群

離線簽名驗證、特徵解耦、局部匹配、Offline Signature Verification、Feature Disentanglement、Local Matching

Using Vision Language Models to Assist Diffusion Models in Generating Realistic Interaction Scenes

智能系統研究所・共同指導：鄭文皇

多媒體圖像處理、文本生成圖像、擴散模型、視覺語言模型、人與物互動生成、Multimedia Image Processing、Text-to-Image Generation、Diffusion Model、Vision Language Models、Human-object Interaction Generation

MMM-TVR: Many-to-Many Matching for Text-Video Retrieval in a Novel Probabilistic Basic Concept Space

電機資訊國際碩士學位學程・共同指導：黃敬群

視覺語言任務、文字影片檢索、表徵學習、半監督對比學習、軟偽標籤、Vision-Language Task、Text-Video Retrieval、Representation Learning、Semi-supervised Contrastive Learning、Soft Pseudo-labels

TrajFine: Predicted Trajectory Refinement for Pedestrian Trajectory Forecasting

電機工程學系

路徑預測、擴散模型、混合資料擴增、課程式學習、深度學習、Trajectory Prediction、Diffusion Model、Mixup Augmentation、Curriculum Learning、Deep Learning

On Recommerder Systems of Next Generation Applications on Social Networks

資訊科學與工程研究所・共同指導：黃俊龍

推薦系統、社群網路、虛擬實境、線上直播、共張量分解、共演化、Recommender System、Social Network、Virtual Reality、Live Stream、Tensor Co-factorization、Co-evolution

Learning Efficient Interaction Anchor for HOI Detection

電機工程學系・共同指導：鄭文皇

人與物交互檢測（HOI）、交互錨點、Transformer、Human-object interaction detection (HOI)、Interaction anchor

Domain Adaptive Edge-guided DETR for Object Detection

電機資訊國際碩士學位學程・共同指導：鄭文皇

DETR、Domain Adaptation、Object Detection、Cityscapes

Scheduling-Sparsity Pruning Approach for Large Language Models

人工智慧技術與應用碩士學位學程・共同指導：鄭文皇

大型語言模型、模型壓縮、模型剪枝、稀疏性、large language model、model compression、model pruning、sparsity

Dual Memory-guided Probabilistic Model for Weakly-supervised Anomaly Detection

智能系統研究所・共同指導：鄭文皇

異常偵測、異常定位、擴散模型、擴散引導、弱監督式學習、記憶庫、anomaly detection、anomaly localization、diffusion model、diffusion guidance、weakly-supervised learning、memory bank

Toward Artifact-free Virtual Try-on via Pre-warping Partitioned Clothing Alignment

電機工程學系

虛擬試穿、深度學習、Virtual try-on、Deep learning

Using Language Model to Label Intent for Improving Hierarchical Table Question Answering

電機工程學系

表格問答、大型語言模型、深度學習、數據分析、Table question answering、large language model、deep learning、data analysis

Controllable Language Generation and Cross-modal Representation Enhancement

電機工程學系

人工智慧、深度學習、自然語言生成、跨模態特徵學習、可控式文字生成、Artificial Intelligence、Deep Learning Language、Generation、Cross-modal Representation Learning、Controllable Text Generation

Low-Resource Learning Methodologies for Vision and Language Tasks

電機工程學系

深度學習、低資源學習、遷移學習、元學習、視覺語言模型、Deep learning、Low-resource learning、Transfer learning、Meta learning、Vision-language models

Multi-Target Video Moment Retrieval by Semantic Fusion Augmentation With Intra-Video Contrastive Loss and Semantic Boundary Detection

電機工程學系

視覺語言任務、多模態學習、資訊檢索、影片片段檢索、跨模態檢索、Vision-Language Task、Multimodal Learning、Information Retrieval、Video Moment Retrieval、Cross-Modal Retrieval

Density-Based Flow Mask Integration via Deformable Convolution for Video People Flux Estimation

人工智慧技術與應用碩士學位學程

人群技術、影像監控、深度學習、Crowd Count、Video Surveillance、Deep Learning

Fast Adaptation in Object Detection Under Adverse Weather via Image Adaptation, Attention Mechanisms, and Meta-Learning

電機工程學系

物件偵測、圖像適應、惡劣天氣、注意力機制、元學習、快速適應、object detection、image adaptive、adverse weather conditions、attention mechanisms、meta-learning、fast adaptation

Fish Localization Using Dual-Frequency Identification Sonar Imaging Videos

電機資訊國際碩士學位學程

雙頻識別聲納數據集、視頻目標檢測、自適應閾值、對比度有限自適應閾值均衡、Dual-frequency Identification Sonar dataset、Video Object Detection、Adaptive threshold、Contrast Limited Adaptive Threshold Equalization

Traffic Events Understanding Based on Closed-Circuit Television Cameras

電機工程學系

異常檢測、閉路監視錄影機、深度學習、Anomaly Detection、Closed-Circuit Television Camera、Deep Learning

Beyond Detection: A Defend-and-Summarize Strategy for Robust and Interpretable Rumor Analysis on Social Media

電機工程學系

謠言檢測、社群網絡分析、對抗式攻擊與防禦、可解釋性、文本摘要、非監督式學習、自監督式學習、rumor detection、social network analysis、adversarial attack and defense、interpretability、text summarization、unsupervised learning、self-supervised learning

Defect Generation with Inpainting Method to Improve Defect Classifier

電機工程學系

自動光學檢測 (AOI)、缺陷生成、自動缺陷檢測、表面缺陷檢測、圖像修復、automated optical inspection (AOI)、defect generation、automated defect detectio、Surface-defect detection、image inpainting

Multi-Speakers Single Channel Speech Separation with Hierarchical Structure

電機工程學系

語音分離、層次結構、單聲道、語音嵌入、分類模型、speech separation、hierarchical structure、single channel、voice embedding、classification model

Prediction of Music Memorability Based on Deep Learning

電機工程學系・共同指導：張文輝

深度學習、音樂記憶性預測、神經網路模型、梅爾頻譜、資料擴增、自監督預訓練音訊頻譜轉換器、SHAP 可解釋性分析、deep learning、music memorability prediction、neural network、mel-spectrum、data augmentation、self-supervised pre-trained audio spectrogram transformer、interpretability analysis ii

Temporal Difference-Aware Graph Convolutional Reinforcement Learning for Multi-Intersection Traffic Signal Control

電機工程學系・共同指導：王蒞君

機器學習、強化學習、交通號誌控制、參數共享、圖卷積神經網路、Machine Learning、Reinforcement Learning、Traffic Signal Control、Parameter Sharing、Graph Convolution Network

Siamese Neural Network-based Cross-Component Continual Learning Defect Detection

電機工程學系

瑕疵偵測、持續學習、孿生神經網路、圖像檢索、深度學習、Defect detection、Continual learning、Siamese neural network、Image retrieval、Deep learning

Improving entity disambiguation using knowledge graph regularization

電機工程學系

自然語言處理、實體辨識、知識圖、natural language processing、entity disambiguation、knowledge graph

Data-efficient Crowd Density Estimation via Spatiotemporal Dilated Convolution and Self-Relabeling

電機工程學系

深度學習、電腦視覺、人群計數、空洞卷積神經網路、自我重新標註、deep learning、computer vision、crowd count estimation、dilated convolutional neural network、self-relabeling

Face-based Voice Conversion: Learning the Voice behind a Face

電機工程學系

語音轉換、零樣本學習、多模態學習、視聽關係、重參數化、voice conversion、zero-shot learning、multimodal learning、audio-visual relationship、reparameterization

A Study of Practical Algorithms and Benchmarks for Network Anomaly Detection

人工智慧技術與應用碩士學位學程・共同指導：鄭文皇

網路異常偵測、入侵檢測資料集、演算法、機器學習、集成學習、神經網路、Network anomaly detection、Network-based intrusion detection dataset、Algorithm、Machine learning、Ensemble learning、Neural network

Imbalanced Aspect Ratio of Surface Defect Detection Using Small Data

電機工程學系・共同指導：王蒞君

物件辨識、卷積神經網絡、瑕疵檢測與分類、自動光學檢測、資料擴增技術、Object Detection、Convolutional Neural Network (CNN)、Defect Inspection and Classification、Automatic Optical Inspection (AOI)、Data Augmentation Technique

On Ranking Multi-source Review Package via Explainable Deep Self-Attentive Network

電機工程學系

多來源評論、推薦系統、深度學習、multi-source reviews、recommendation system、deep learning

Personalized Subject Generation via User-Subject Tensor Co-Factorized Embeddings

電信工程研究所・共同指導：張文輝

自然語言處理、推薦系統、深度學習、Natural Language Processing、Recommendation Systems、Deep Learning

Zooming In for Text Detection with Blurry Text

電機工程學系

場景文本偵測、模糊文字、Scene Text Detection、Blurry Text

A 307 GOPs CNN Accelerator for 30 fps Object Detection SoC using a 28nm CMOS technology

電機工程學系・共同指導：闕河鳴

物體辨識、類神經網路、硬體加速器、影像辨識、機器學習、YOLO、Object Detection、Convolution Neural Network、Hardware Accelerator、Machine Learning

Domain-Adaptive Object Detection via Uncertainty-Aware Distribution Alignment

電機資訊國際學程

域適應、物件偵測、Domain Adaptation、Adversarial Learning、Uncertainty-Aware、Distribution Alignment、Object Detection

Learning Visual Object Tracking using Small Data

電機工程學系

視覺追蹤、電腦視覺、自監督學習、少樣本學習、卷積神經網路、Visual Object Tracking、Computer Vision、Self-supervised Learning、Small Data、Convolutional Neural Network

Reconstructing QRS Complex from PPG by Transformed Attentional Neural Networks

電機工程學系

卷積神經網路、心電圖、編碼器-解碼器、光體積變化描記圖法、轉換網路、convolutional neural network、electrocardiography、encoder-decoder、photoplethysmography、transform network

Explainable Machine Health State Prediction through Multi-Channel Attention

電機學院電信學程

機器健康狀態預測、通道注意力、自我注意力、可解釋性、machine health state prediction、channel attention、self-attention、explainability

Neuron Differentiated Network Expansion via Approximated Fisher-Rao Metric

電機工程學系

機器學習、深度學習、網路擴增、功能性保留轉換、費雪-拉奧範數、知識轉移、Machine Learning、Deep Learning、Network Expansion、Functional Preserving Transformation、Fisher-Rao Norm、Knowledge Transfer

Autonomous Navigating to Target with Collision Avoidance for Unmanned Aerial Vehicles

電機工程學系・共同指導：王蒞君

Reinforcement Learning、Deep Learning、Machine Learning、Motion and Path Planning、Applications of Reinforcement Learning、強化式學習、深度學習、機器學習、路徑規劃

On Accelerating Multi-Layered Heterogeneous Network Representation Learning via Landmark Selection

電機工程學系

異質表示學習、網路表示、特徵學習、維度縮減、元路徑、地標選擇、表徵學習、Heterogeneous representation learning、Network representation、Feature learning、Dimension reduction、Metapath、Landmark selection、Embedding learning