Carbonoi founder co-develops Deep Learning AI to sort waste and reduce urban flooding

Our co-founder, Associate Professor Thanarat Chalidabhongse (Ajarn Fon), has successfully co-developed AI technology to detect waste in canals. 🎉Published News: https://www.bangkokbiznews.com/finance/stock/1147053

Problem Statement

Thailand ranks among the top 10 countries for dumping waste into the ocean. Many may not realize that the trash we dispose of daily gets washed into canals, rivers, and ultimately, the sea. The Chao Phraya River was selected by OCEAN CLEANUP, a global plastic removal organization, for a three-year study of waste flow due to the dense riverside communities. This trash not only contributes to marine pollution but also causes urban flash floods by obstructing waterways, which is why we must identify the sources to reduce waste effectively.

Innovation

AI, using Deep Learning, has become a solution to help tackle this issue. Developed by a Thai engineering team from Chulalongkorn University, AI systems were trained with over ten thousand images to categorize waste types. The results reveal a direct correlation between heavy rainfall and increased waste in the canals.

The development of AI to detect and track objects from moving images marks a significant advancement in addressing two challenges:

1. Object Detection – AI identifies specific objects in the images and classifies them accurately.

2. Object Tracking – AI tracks the movement of objects across video frames to count them without duplicating.

   
   

To enhance accuracy, the AI needs millions of images for learning, and high-quality hardware, such as expensive cameras, can capture faster, clearer images.

Behind-the-Scene

Unexpectedly, Ajarn Fon encountered an issue when the international cameras used recorded footage to the manufacturer’s server before downloading it for local analysis. The team adapted the software to retrieve images directly from the cameras.

Having local developed hardware and software would reduce risks of data leakage and costs, making this AI vision technology a crucial tool in monitoring natural disasters and other environmental challenges across many dimensions.