Research fields at AImageLab

Videosurveillance & HBU

Learning to Detect and Track Visible and Occluded Body Joints in a Virtual World GAN4Surveillance: Generative Adversarial Networks for Attribute Classification Duke Imagelab Multi-Target, Multi-Camera Tracking Project Spotting Prejudice Action and Gesture Recognition for Human Computer Interaction Group Detection and Crowd Analysis Multiple People Tracking People Re-identification ViSOR - Video Surveillance Online Repository People trajectory analysis and anomaly detection People Tracking From Multiple Cameras

Computer Vision & Pattern Recognition

Visual Saliency Prediction Connected Components Labeling Animal welfare analysis from 3D sensors 3D Computer Vision 3D Reconstruction of Skin Lesions for Tumor Diagnosis in OCT imaging

Multimedia & Big Visual Data

Video Captioning with Naming Scene detection in Broadcast Videos Deep Learning in videos Video Captioning Class Specific Segmentation Garment Selection and Color Classification Egocentric Video Summarization Face Recognition in News Streams

Cultural Heritage and Digital Humanties

Art2real: translating artworks to photo-realistic images Visual-semantic domain adaptation in Digital Humanities Layout and content analysis in Digitized Books Egocentric Video Registration and Architectural Details Retrieval EgoVision and Human Augmentation for Cultural Heritage

Automotive

Learning to Generate Faces from RGB and Depth data Face Verification with Depth Images Hand Monitoring and Gesture Recognition for Human-Car Interaction Landmark Localization in Depth Images Driver Attention through Head Localization and Pose Estimation Dr(eye)ve a Dataset for Attention-Based Tasks with Applications to Autonomous Driving Learning to Map Vehicles into Bird Novelty & Anomaly detection

New Visions: sensors, mobile and embedding

Egocentric Vision for Detecting Social Relationships Sensing floors Real Time Ellipse Detection on Mobile Devices Collaborative robot programming