Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/universaldatatool/coronavirus-mask-image-dataset
Image dataset from Instagram of people wearing medical masks, no mask, or a non-medical (DIY) mask
https://github.com/universaldatatool/coronavirus-mask-image-dataset
computer-vision coronavirus covid-19 covid19 dataset fastai images machine-learning
Last synced: 4 days ago
JSON representation
Image dataset from Instagram of people wearing medical masks, no mask, or a non-medical (DIY) mask
- Host: GitHub
- URL: https://github.com/universaldatatool/coronavirus-mask-image-dataset
- Owner: UniversalDataTool
- License: mit
- Created: 2020-04-09T14:58:55.000Z (over 4 years ago)
- Default Branch: master
- Last Pushed: 2020-05-19T20:05:57.000Z (over 4 years ago)
- Last Synced: 2024-10-29T18:41:50.404Z (22 days ago)
- Topics: computer-vision, coronavirus, covid-19, covid19, dataset, fastai, images, machine-learning
- Language: Jupyter Notebook
- Size: 2.62 MB
- Stars: 55
- Watchers: 5
- Forks: 17
- Open Issues: 4
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# COVID19 Mask Image Dataset
## Note: Instagram has expired the URLs in this repo.
> Note: This dataset links to images on Instagram. We do not store or own the images on Instagram.
Image dataset from Instagram of people wearing medical masks, non-medical (DIY) masks, or no mask. Created using the [Universal Data Tool](https://github.com/UniversalDataTool/universal-data-tool) for helping people come up with creative solutions for COVID-19 problems. The dataset currently has roughly ~1,205 image samples.
This dataset could be used to build a face mask detector for selfie-type photos.
## Getting the Data
[Click this to download mask_labels.udt.csv](https://github.com/UniversalDataTool/coronavirus-mask-image-dataset/raw/master/mask_labels.udt.csv). (you might have to "Save Page As" to save it)
It looks kinda like this:
| path | . | imageUrl | output |
| --------- | - | ------------------------------ | ---------------- |
| samples.0 | . | https://example.com/image1.jpg | medical_mask |
| samples.1 | . | https://example.com/image2.jpg | no_mask |
| samples.2 | . | https://example.com/image3.jpg | non_medical_mask |
## How to View
Download the [mask_labels.udt.csv](https://github.com/UniversalDataTool/coronavirus-mask-image-dataset/raw/master/mask_labels.udt.csv) file and open it at [universaldatatool.com](https://universaldatatool.com)
## How to Use
[Check out this notebook](https://github.com/UniversalDataTool/coronavirus-mask-image-dataset/blob/master/FastAI%20Classification%20Model.ipynb) to see how to use the dataset with [fast.ai](https://fast.ai). You could also download the [mask_labels.udt.csv](https://github.com/UniversalDataTool/coronavirus-mask-image-dataset/blob/master/mask_labels.udt.csv) file to use whatever framework you want.
We were able to achieve 93% accuracy prediction accuracy on the dataset (see confusion matrix below):
