New Trash Classification Dataset

Structured visual dataset for training recyclable object classification models, with 1,000 images per class for 8 different categories.

Download dataset

Size

8,000 JPG images divided into 8 classes

Licence

CC BY 4.0

Description

‍

The dataset New Trash Classification includes 8,000 images divided equally into 8 waste classes (e.g. plastic, metal, paper, organic). Each image has been selected and restructured from several open-source sources to offer a clean, balanced and relevant corpus for training deep learning models.

‍

What is this dataset for?

‍

Forming image classification models in an environmental context
AI projects to automatically sort recyclable waste
Mobile or industrial applications to help with selective sorting

‍

Can it be enriched or improved?

‍

Yes, you can increase the diversity of the dataset via data augmentation, add subclasses (crumpled paper, soft/hard plastic...), or combine with geolocated data to enrich use cases. It can also be used as a basis for multimodal approaches (image + text).

‍

🔎 In summary

Criterion	Evaluation
🧩 Ease of use	⭐⭐⭐⭐⭐ (Well-structured dataset by folder)
🧼 Need for cleaning	⭐⭐⭐⭐⭐ (Low – content already filtered)
🏷️ Annotation richness	⭐⭐✩✩✩ (Limited to class labels)
📜 Commercial license	✅ Yes (CC BY 4.0)
👨‍💻 Beginner friendly	🌟 Yes, perfect for initial CNN models
🔁 Fine-tuning ready	🎯 Useful as a base for pre-trained models
🌍 Cultural diversity	🔹 Neutral – universal everyday objects