Teach the AI! A Participatory Technology Development Trial: Helping Everyone Grow a Capybara AI Behavior Analysis Tool

AQUARIUM×ART átoa (hereinafter "átoa") is conducting a trial exhibit for a "proprietary tool that automatically records and analyzes the behavior of captive animals using image AI" (hereinafter "this tool"), which is being co-developed with Kobe Digital Labo, Inc. (Chuo-ku, Kobe, Hyogo Prefecture; Shinichi Tamaki, President; hereinafter "KDL").

This tool is being developed to objectively evaluate the effectiveness of environmental enrichment(*) implemented by átoa. A major feature of this exhibit is that it is participatory; as visitors engage with it, the AI grows, providing an interactive learning experience where visitors can naturally enjoy learning about animal behavior and ecology.

Through this trial exhibit, we plan to verify the usability and necessary functions for visitors, and provide feedback for development toward full-scale exhibition and operation.

[Image: Participation concept] [Image: átoa's capybaras]

Exhibit Overview Now that the prototype of this tool is complete, we are conducting a trial exhibit featuring capybaras. Visitors can use a tablet installed at the exhibit corner to draw a box around the capybara in the image (a process called annotation) and select (record) its behavior at that time. As the input image data increases to 100, 200, 500, and 1,000 images, the AI's ability to determine when, where, and what the capybara is doing from the images improves. By using this AI analysis tool nurtured by visitors, it becomes possible to track capybara behavior in detail over long periods, creating a system that allows for objective evaluation of which environmental enrichment methods are effective.

[Image: iPad input screen] [Image: Accumulated data]

[Development Objectives]

① To objectively evaluate environmental enrichment Environmental enrichment is an initiative to devise housing environments and management methods to bring out an animal's natural traits and increase their choices; it is one of the concrete methods for improving animal welfare. What is important is not to let it end in mere self-satisfaction, but to objectively evaluate whether those efforts are truly benefiting the animals. Only by correctly judging effectiveness through evaluation and repeating improvements can we lead to a better life for the animals. On the other hand, recording behavior requires someone to observe constantly or to watch videos repeatedly to take notes. If this AI analysis tool is nurtured, the time constraints for observation will be eliminated, and objective, detailed evaluation will allow for the selection of effective environmental enrichment initiatives.

② To conduct research and create exhibits together with our guests Most exhibits at aquariums and zoos are things to be viewed. While there are some things to experience, there have been almost no exhibits that are created together. With the thought that people might become more deeply interested in living creatures by nurturing rather than just experiencing, and by experiencing rather than just watching, we wanted to create an exhibit that we build together with our guests. Improving the analysis accuracy for creatures with little accumulated data requires many opportunities for AI learning. Through participation in the development of this tool, we hope that guests will also be able to touch upon the charm of living creatures more deeply.

③ To cultivate local IT talent in Kobe This AI behavior analysis tool was created under the guidance of KDL by interns accepted by KDL as part of an initiative to foster local IT talent and promote employment in Kobe. As a unique and attractive internship project theme for the IT field, they have been working on it energetically.

[Implementation Overview] Exhibit Period: April 1, 2026 (Wed) – April 17, 2026 (Fri) Exhibit Location: átoa 4F SKYSHORE Participation Fee: Free. *Separate admission fee required. * Please wait your turn and share the equipment. * Exhibit content is subject to change.

■ Contact for Kobe Digital Labo [Image: Contact information] Kobe Digital Labo, Inc. Digital Business Division, Data Intelligence Team Kohei Yamaguchi, Toshiki Haraguchi, Taichi Fukuoka E-mail: info@kdl.co.jp / Phone: 078-327-2280

<Basic Information> - Location: 7-2 Shinko-cho, Chuo-ku, Kobe, Hyogo 650-0041 - Access: Approx. 18-minute walk south from Sannomiya Station (approx. 1.4 km) - Website: https://atoa-kobe.jp Operating Company: Aquament Inc. * Please check the official website for the latest business hours.