SenseCollect: We Need Efficient Ways to Collect On-body Sensor-based Human Activity Data!
summary
SenseCollect implements a smartwatch app and laptop UI, runs control-variable experiments, and produces 240+ subjects across deployments to test collection protocols. Key claims: monitoring by researchers is unnecessary, paying for data quality and quantity yields usable recordings, and real-time sensing feedback (visual traces on laptop/smartglasses) plus breaks and music improve quality. The team collects 192 hours from 32 participants in one week and then 113 participants in a follow-up, and releases the resulting datasets.
AI-written summary · about
related talks
28mDiscussion: Humans As Sensors
44mQuantified Self: Analyzing the Big Data of our Daily Life
05mCollecting Massive Data via Crowdsourcing
35mData collection from embedded devices: challenges and solutions
34mMining smartphone sensor data with python
01h 04mTake Control of the Data of You
06mNeckSense: A Multi-Sensor Necklace for Detecting Eating Activities in Free-Living Conditions
09mAn Activity Recognition System for Taking Medicine Using In-The-Wild Data to Promote Medication ...