Collection and annotation of
Litter dataset

"Litter consists of waste products that have been discarded incorrectly, without consent, at an unsuitable location."
Everyone realizes how litter can harm our environment. Computer vision and object detection applications can help. The abundance of security cameras in cities and your backyards can not only find garbage on the street but also detect litter being thrown. Automatic waste sorting machines are used to sort different types of waste.
The recycling company was interested in dataset collection and annotation of waste images. Garbage should be on the street, lawn, glade as passers-by left it. Litter should be labeled with bounding boxes and belongs to one of four categories: Glass, PET, Aluminum, and Beverage Carton. No "Biodegradable waste" and "Non-recyclable waste".

We collected the dataset by taking pictures of litter in the streets, lawns, and parks in nine cities of four countries.