| --- |
| task_categories: |
| - image-classification |
| tags: |
| - waste |
| - classification |
| pretty_name: waste-cl |
| --- |
| |
| # Dataset Card for waste classifier |
|
|
| This dataset contains waste images in different categories: |
| - cardboard |
| - compost |
| - glass |
| - metal |
| - paper |
| - plastic |
| - trash |
|
|
|
|
| ### Dataset Description |
| - **Curated by:** Rootstrap |
| - **License:** MIT |
|
|
| ### Dataset Sources |
| Data is a combination of [Trashnet](https://github.com/garythung/trashnet) dataset plus more images obtained by internet search. |
| Paper: [Classification of Trash for Recyclability Status](https://cs229.stanford.edu/proj2016/report/ThungYang-ClassificationOfTrashForRecyclabilityStatus-report.pdf) |
|
|
| ## Uses |
| The dataset can be used for waste classification or other type of project. |
|
|
| ### Direct Use |
| This dataset is used to build a waste classifier for categorizing different types of waste, being able to correctly throw the trash in the corresponding trash can at our office. |
|
|
| {{ direct_use | default("[More Information Needed]", true)}} |
| |
| ## Dataset Structure |
| The data is already split in train and test folders. |
| Inside each folder contains one folder for each class. |
| |
| ## Dataset Creation |
| |
| ### Curation Rationale |
| |
| at Rootstrap, our Machine Learning Engineers are committed to creating awareness of correct waste classification to help the environment. |
| Their determination to make an impact led to the creation of 'RootTrash', an internal AI-powered app to help us recycle correctly. |
| |
| |
| #### Data Collection and Processing |
| Some of the images were obtained using Bing searcher using the api HTTP. |
| You can find the code used to download the images at this [Google Colab](https://colab.research.google.com/drive/1JvAYFx1DIEi1MMyI-tuCfE2eHMSKisKT?usp=sharing). |
| |
| #### Who are the source data producers? |
| Thung, G., & Yang, M. (2016). Classification of Trash for Recyclability Status. |
| |
| |
| |
| ## Bias, Risks, and Limitations |
| Current model has been trained mostly with internet images and most of them has white background. This might be an issue when testing with real images. |
| In the future, the dataset will be extended with the photos taken through the app. |
| |
| ### Recommendations |
| Integrate this model with a detection model such as [rootstrap-org/waste-detector](https://huggingface.co/rootstrap-org/waste-detector) |
| |
| |
