Visual Plastic Type Recognition Dataset
Dataset of real images classified by type of plastic (PET, PE, PP, PP, PS, PC, Others) for projects in AI, sorting and recycling.
Over 3,000 JPEG images arranged in 6 folders (PET, PE, PC, PP, PS, Others)
Creative Commons Attribution 4.0 International (CC BY 4.0)
Description
This dataset contains more than 3,000 images classified by type of plastic: PET, PE, PP, PC, PS and others. The images come from a variety of sources such as Unsplash and web scraping tools. It is an ideal working base for projects related to recycling, environmental sustainability and the automatic classification of waste.
What is this dataset for?
- Train image classification models to recognize different plastics
- Improving automatic sorting systems in recycling chains
- Raise public awareness about types of plastic and their recyclability
Can it be enriched or improved?
Yes, the dataset can be cleaned, balanced, or enriched with additional annotations. Increasing data or adding visual contexts (e.g. composite objects) can improve their usefulness for more complex models.
🔎 In summary
🧠 Recommended for
- Computer vision students
- AI projects in the environment
- Recycling startups
🔧 Compatible tools
- TensorFlow
- PyTorch
- Keras
- FastAI
- LabelImg
💡 Tip
Start by cleaning and balancing classes before testing different classification models.
Frequently Asked Questions
What types of plastic are included?
The dataset contains the following categories: PET, PE, PC, PC, PP, PS, and Others (including PVC, ABS, and unclassified objects).
Can this dataset be used in an automated waste sorting project?
Yes, it was designed to train image classification models for the recognition of plastic types.
Should data be cleaned or rebalanced before use?
Yes, the dataset is raw and unbalanced. It is recommended that it be pre-treated for optimal machine learning performance.
