Swiss researchers have developed a system that allows measuring the cleanliness of cities with objective criteria. For the first time, the municipal authorities will have access to scientific data to organize urban cleaning, a sector whose budget is millions of euros in many capitals.
The system works with cameras attached to the vehicles that record the streets of a city. The images are analyzed by a computer program capable of locating objects thrown on the street, of recognizing their nature and classifying them, all at high speed.
The system allows to establish an accurate diagnosis of the streets of a city, determining its level of cleanliness, what waste is found where, what are the most affected areas and at what time of the day there is the largest accumulation of waste.
With this information, municipal cleaning services can plan their work schedule, make the best decisions regarding the machines to be assigned to each site, the frequency with which a district has to be cleaned and even the areas to be cleaned. more assiduously by the nature of the waste, in case they are dangerous for the public health. The system has been described in the journal Computer Vision Systems.
Transcend human perceptions
In the absence of measurable scientific criteria, at present the cleaning services of cities are organized based on human perceptions. They lack tools to proceed with objective measures, the researchers said in a statement .
Each city maintains its cleaning system according to personal strategies, which can be more or less successful. They give as an example the city of Zurich, which sends many times a year to numerous people to walk the streets to assess the waste that is found (bottles, butts …) and establish a diagnosis of the state of the city. Other cities rely on citizens to have this vision.
The new technology is based on vehicles equipped with high resolution cameras, which are the ones that cross the streets visually sweeping everything that there is and detecting abandoned or accumulated objects.
These images are analyzed by a computer, programmed to automatically detect each of the objects filmed and decide if it corresponds to the 25 categories classified as waste (newspapers, butts, leaves of trees …). Then, the program classifies them in the corresponding category.