Activity recognition is the task of classifying daily life activities performed by a human on sensor observations. Using mobile phones accelerometer for this observation is one of the cheapest and most convenient method. Activity recognition is specially gaining attention in care facilities for remote monitoring of daily activities for elders living alone and automatic record creation for nurses in hospitals with a goal of providing better care to patients. One challenge for these applications is bridging the gap between models created within the lab and models for real life.
In this challenge, participants will create methods for recognizing 12 activities conducted by nurses in both lab and in real-life settings. In real world, there is high chance of getting missing labels during experiments since users are busy at work. The main challenge is leveraging laboratory data to improve models in real-life. Training data consists of data collected in both settings but test data consists of data only from the real field (same users as training data).
Download the data from IEEE Dataport here
The evaluation of submissions will be done solely based on the accuracy.
