AgaMiko/waste-datasets-review
List of image datasets with any kind of litter, garbage, waste and trash
What it solves
Researchers and developers who want to train AI systems to detect waste in photos or videos often struggle to find good training data. Waste appears in many different settings — streets, beaches, rivers, underwater, drone shots, recycling plants — and each setting requires differently labeled images. This project solves that by collecting and organizing dozens of public waste-image datasets into one central list, so people can quickly pick the right one for their needs.
How it works
The repository is a structured review of datasets. It groups them by where the images were taken (e.g., streets, beaches, underwater, rivers, drone, recycling plants) and by the type of annotation each dataset provides: classification (what is in the image), detection (where objects are), or segmentation (pixel-level outlines). A summary table lists every dataset with its number of categories, number of images, annotation type, and license. Each dataset also has its own short description with a link to its source. The README notes updates, and an annotation-types figure visually explains the difference between classification, detection, and segmentation.
Who it’s for
This is for AI researchers, machine learning engineers, and hobbyists who want to build models that identify or locate litter, garbage, or waste in images and video. It is especially useful for those working on environmental cleanup projects or computer vision tasks where waste-specific training data is required.
Highlights
- Over 40 datasets are listed, covering diverse environments: beaches, underwater, rivers, drone/satellite, recycling plants, and more.
- Datasets are grouped by setting and annotation type, making it easy to find a matching dataset for a specific project.
- Each entry includes size, annotation format, license, and a direct link (GitHub, Zenodo, Kaggle, etc.).
- The project was created as part of the Detect Waste initiative, which aims to use AI to detect plastic waste in the environment.
- The list is actively maintained — a recent update added 16 new datasets and fixed broken links.
Related
- Project
- Project
- Project
- Project