A research group from Kagoshima University Graduate School, the Japan Agency for Marine-Earth Science and Technology, and Kyushu University has developed a system that visualizes the amount of garbage in cities by type using "Pirika," a garbage-picking SNS developed by Pirika Inc., and an image analysis AI based on deep learning.
It is said that the majority of marine plastic pollution (around 8%) is household waste that has flowed into cities. In particular, a lot of household waste washes up on the coast near rivers. To reduce plastic pollution, it is extremely important to visualize where, how much, and what type of waste is polluting the city, which is the source of pollution.
The smartphone app for picking up trash, Pirika, is used in 132 countries around the world, and has picked up a total of 3.6 million pieces of trash. The newly developed system combines this smartphone app with deep learning. This makes it possible to visualize trash in the city, identify the items and locations that are sources of pollution, and contribute to the planning of prioritized pollution countermeasures.
In addition, it was unclear to what extent street cleaning activities contribute to the sustainable beautification of the city. If this technology can be used to establish a regular observation system for a specific area, it will be possible to visualize the extent to which the cleaning effect contributes to the subsequent increase in the amount of garbage (for example, whether or not the increase in garbage is suppressed according to the broken windows theory).
Similarly, if measures to reduce emissions of a certain item are implemented based on the results of such regular observations, the effectiveness of those measures can be visualized on a map and made public. The key to making this a reality is citizen science, and it is hoped that data collection through smartphone apps with the participation of citizens will lead to the development of a system specialized for each region.
Paper information:[Waste Management] Quantification of litter in cities using a smartphone application and citizen science in conjunction with deep learning-based image processing
