Automated Analysis of Nursery School Observations

Automated Analysis of Nursery School Observations

Jien Kato (Nagoya University, Japan)
DOI: 10.4018/978-1-4666-2196-1.ch029


This chapter introduces an ongoing project with the goal of measuring and analyzing children’s behavior automatically. Some key technologies, including methodologies for acquiring data, tracking a target across different cameras over time, identifying individuals, activity recognition, interaction analysis, and behavior summarization for a target child are presented. Some encouraging results from a real system developed in a nursery school environment are also described. As these technologies enable the content-based retrieval, comparison, and summarization of large-scale observational data, they are applicable to various purposes, such as healthcare, diagnosis, and the assessment of children’s development.
Chapter Preview

Core Technological Problems

In the MIMAMORI project, we record children’s daily lives by the use of video, audio and temporal-spatial tags of moving targets. However, to protect privacy, we have not yet utilized the audio portion. As described above, our goal is to obtain an automated measurement and analysis of children’s behavior to enable the content-based retrieval, comparison, and summarization of large-scale observational data. To achieve this goal, we face the following technological problems:

  • Tracking a target child across multiple cameras over time.

  • Identifying individual children across populations.

  • Recognizing a target child’s activities.

  • Analyzing a target child’s interactions with people.

These technologies are applicable to various purposes, such as parenting support, healthcare, and diagnosis, and are useful not only for the assessment of children but also for the elderly and other populations.

System Setting

In this work, the observational data of children’s behavior are acquired in a noninvasive manner. We set up a data-sensing system in the nursery school within Nagoya University campus, which consists of seven cameras, each integrated with a RFID (Radio Frequency Identification) receiver. All of the seven camera-receiver pairs are installed in places where the children regularly appear: one in the entrance, four in the nursing rooms, one in the resting room and one in the passageway, as shown in Figure 1. Data recording from these seven locations occurs ten hours a day throughout the year.

Figure 1.

Hardware configuration for data sensing

Each camera is placed adjacent to an RFID receiver, which catches the signals sent by nearby RFID tags. Upon entering the nursery school each morning, every child is given an RFID tag to keep in his/her pocket. Subsequently, the tags will continuously send out the holder’s ID, and the nearby receivers can capture the IDs and write them into a log file. This log information will later be used to effectively pick out videos that are expected to contain a specific child (called target-specific videos).

Complete Chapter List

Search this Book: