Evaluating the Use of a Dialogue System For Activity Data Collection
Tittaya Mairittha, Nattaya Mairittha, Sozo Inoue,
The annotation of human activity is the essential process for human activity recognition system. The problem was that annotation systems require cumbersome equipment which deprives their use. This paper presents a method to collect training labels for human activity recognition by using a dialogue system. The experiments were carried to predict users' activity and provided valuable insights on how users interact with the system. The evaluation shows that the dialogue system can collect data efficiently by achieving the highest precision, recall, and F1 score is 0.78 and 0.75 and 0.76, respectively.