InfiniTouch: Finger-Aware Interaction on Fully Touch Sensitive Smartphones
summary
A phone prototype uses two Nexus 5 units and an Arduino with capacitive sensors on front, back and three sides to capture low‑res “capacitive images”; a motion‑capture study with 9 million capacitive frames from 20 participants trains a CNN that identifies fingers at 95.78% and estimates 3D fingertip position with mean absolute error ≈0.74 cm (X/Y 0.85 cm, Z 0.53 cm). On‑device inference runs ≈24.7 ms (tracking alternative 3.4 ms). Demonstrations show finger‑specific gestures and flexion as a modifier work reliably; schematics, code and dataset are released.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 2:36Sample Use Case
- 4:03Hardware Overview
- 4:35Handheld Device
- 5:53Capacitive Image
- 6:32Methodology
- 7:57Finger positions
- 8:13Preprocessing
- 9:06Model Development
- 9:54Loss Function
- 10:30CNN Architecture
- 11:50Deployment
- 12:19Tracking
- 12:55Future Work
- 14:10Summary
related talks
16mInvestigating the feasibility of finger identification on capacitive touchscreens using deep ...
18mEstimating the Finger Orientation on Capacitive Touchscreens Using Convolutional Neural Networks
14mDeepFisheye: Near-Surface Multi-Finger Tracking Technology Using Fisheye Camera
17mPredicTouch: A System to Reduce Touchscreen Latency using Neural Networks and Inertial ...
19mBackHand: Sensing Hand Gestures via Back of the Hand
19mATK: Enabling Ten-Finger Freehand Typing in Air Based on 3D Hand Tracking Data
14mLeveraging Dual-Observable Input for Fine-Grained Thumb Interaction Using Forearm EMG
12mEclipseTouch: Touch Segmentation on Ad Hoc Surfaces using Worn Infrared Shadow Casting