Moreover, body movements are incorporated as part 
of the observation behavior of a fossil exhibition, 
which previously consisted of a conventional written 
explanation. The system enables learners to enhance 
their sense of immersion in a paleontological 
environment and learn about the fossil itself and its 
paleontology. 
In this paper, we summarize the prototype of 
"BELONG" as the first step toward developing the 
immersive learning support system for the fossil 
exhibition at the museum. In addition, we describe the 
results of our experimental evaluation of the learning 
support and immersion abilities of the system with the 
aim of clarifying whether it can provide learners with 
a realistic paleontological observation experience. 
2  LEARNING SUPPORT SYSTEM 
2.1 Belong 
We aim to realize the immersive learning support 
system "BELONG" that simulates a paleontological 
environment and transitions that are impossible to 
experience in reality for efficient learning at the 
museum. Figure 1 illustrates the concept of 
"BELONG." This system accepts body movements as 
input for observational behavior. The movements of 
the whole body and the system operation are linked; 
therefore, it is possible to enhance the sense of 
immersion in the paleontological environment. The 
sense of immersion improves if the system can be 
operated in conjunction with complicated body 
movements as compared with a case in which the 
system is operated with simple body movements. The 
recognition of complicated body movements should 
not involve attaching expensive sensors or devices to 
learners when implementing it in a museum. In this 
system, we utilize Microsoft’s Kinect v2 sensor, a 
range-image sensor originally developed as a home 
videogame device. Because BELONG comprises 
only a Kinect v2 sensor, projector, and control PC, it 
allows us to provide a low-cost immersive learning 
experience within a small space. The advantage of 
this arrangement is that it is possible to easily change 
the learning contents. Moreover, we recognize the 
body movements of learners by gesture recognition 
using the Kinect v2 sensor. The gesture recognition 
system, which can also interpret complicated body 
movements, registers the body movement the creator 
wishes to recognize and judges whether it is 
recognized by verifying the similarity with the body 
movement.   
 
Figure 1: Concept of BELONG. 
2.2  Configuration of the System 
We developed an immersive learning support system 
"BELONG" that simulates a paleontological 
environment and transitions that are impossible to 
experience in reality.
  As a first step towards the 
realization of this system, we are developing a system 
to simulate paleoecology, especially learning about 
dinosaurs, based on experiences that simulate a 
paleontological excavation.
 Our assumption was that 
learners' interest would increase by virtually 
excavating fossils included in the current exhibition. 
However, because excavation motions are complex 
body movements, gesture recognition was used. 
(Tokuoka, M., 2017) When the excavation proceeds 
successfully, videos showing the characteristics of the 
dinosaur are displayed.
  Linking the body and the 
video in this way increases the sense of immersion. 
These body movements are recognized by a Kinect v2 
sensor, the properties of which are described below. 
Microsoft’s Kinect v2 sensor is a range-image 
sensor originally developed as a home videogame 
device. Although it is inexpensive, the sensor can 
record sophisticated measurements regarding the 
user’s location. Additionally, this sensor can 
recognize humans and the human skeleton using the 
library in Kinect’s software development kit for 
Windows. Kinect can measure the location of human 
body parts such as hands and legs, and it can identify 
the user’s pose or status with this function and the 
location information. Moreover, Kinect Studio and 
Visual Gesture Builder are used to recognize 
complicated body movements captured by the Kinect 
sensor. These enable complicated body movements to 
be recognized using the discriminator (Tokuoka, M., 
2017). By using these, it is possible to create a 
discriminator that registers the body movements we 
want to recognize and can accurately recognize body 
movements using machine learning. As a complicated