GO-Finder: A Registration-Free Wearable System for Assisting Users in Finding Lost Objects via ...
summary
A wearable neck camera system automatically detects hands, crops held-item patches, and clusters instances using convolutional neural network features with cosine-similarity matching to produce a hand-held object timeline of last-seen thumbnails. It requires no pre-registration (registration-free), stores only last frames per cluster to reduce data and privacy risk, and runs on video from a neck-mounted camera. A 12-participant hide-and-seek study shows higher retrieval accuracy and lower mental load than frame-based browsing, with significant improvement over frame-based and large effect sizes versus no aid.
AI-written summary · about
chapters (5)
related talks
14mAssisting group activity analysis through hand detection and identification in multiple ...
19mEveryday Eye Contact Detection Using Unsupervised Gaze Target Discovery
12mInvisibility Cloak: Personalized Smartwatch-Guided Camera Obfuscation
14mOddEyeCam: A Sensing Technique for Body-Centric Peephole Interaction using WFoV RGB and NFoV ...
10mREGROUP: A Robot-Centric Group Detection and Tracking System
47mHow a Tiny Head Detector Powers Intelligent Video Conferencing
40mNew Face, Who Dis? Protecting Privacy in a World of Surveillance
14mBack-Hand-Pose: 3D Hand Pose Estimation for a Wrist-worn Camera via Dorsum Deformation Network