Ikshana: computer vision for street cleaning

At home, we take great care to keep things clean. However, whoever manages the cleanliness of a city needs a more organized approach to monitor every corner that needs cleaning. The Ikshana trash detection system we developed follows a similar idea. It notices unattended garbage and dirty roads, and it applies a better disposal strategy.

The client provides a public road cleaning facility and runs garbage vehicles all over the city. However, the drivers usually ignore the roadside garbage. They collect only the bags that households prepare with their own waste. As a result, a wide range of areas stayed unclean because of the many unclaimed litters. This waste leads to greenhouse gas emissions, which raise the carbon dioxide in the atmosphere and heat up our environment.

Our client also focused on keeping the city and streets clean, so we designed an application named Ikshana. The application has three ends: the admin who manages the dashboard, the camera assignee and the cleaner.

First, every garbage-picking vehicle of the client gets a camera with a unique ID. The database saves this ID with the driver’s name, the vehicle number, and other necessary details. Next, we built the Camera Assignee application with Flutter, and it is server-based. It also identifies objects on the road that the open detection system looks for, for example a leaking dust bin that spills waste, moving animals, filthy water on the road, or a dump. Thus, if the application identifies any of these, it automatically uploads pictures along with the place where it took them, including the latitude and longitude. Then the system automatically generates a ticket, based on the vehicle that made the catch and the person in charge of that vehicle.

Now comes the admin, the authorized person who can view the generated tickets in the administrator dashboard. The admin can appoint the right person to clean the waste in the area, mark any trash as unhandled, and add comments if necessary. The admin also sees the full status of the ticket. Besides, the admin can close or reopen it if the work still needs improvement.

The cleaner is the person to whom the admin assigns the task of cleaning an area. As soon as the ticket changes to in progress, the cleaner can add remarks and photographs to the status update. Finally, the application lets the cleaner change the status to approval after attaching an image and a note as proof. Our AI product engineering team builds solutions like this.

The Ikshana trash detection system became an environment-friendly solution. It streamlined the whole process of cleaning the streets and covered areas that were once left untouched. The dashboard kept the admin, camera assignee, and cleaner on the same page, so it reduced the need for manual inspection. Auto-detection of images that contain trash, and auto-filling of the camera assignee details, saved time that people spent on these tasks by hand. Thus, it promoted quick action.