https://github.com/memudualimatou/spot-mask-challenge
https://github.com/memudualimatou/spot-mask-challenge
argument-parser keras opencv python tensorflow transfer-learning
Last synced: about 1 month ago
JSON representation
- Host: GitHub
- URL: https://github.com/memudualimatou/spot-mask-challenge
- Owner: memudualimatou
- Created: 2020-09-14T09:40:56.000Z (about 5 years ago)
- Default Branch: master
- Last Pushed: 2022-04-26T14:44:23.000Z (over 3 years ago)
- Last Synced: 2025-04-05T20:33:10.715Z (6 months ago)
- Topics: argument-parser, keras, opencv, python, tensorflow, transfer-learning
- Language: Jupyter Notebook
- Homepage:
- Size: 8.05 MB
- Stars: 3
- Watchers: 1
- Forks: 2
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# SPOT-MASK-CHALLENGE
***A Face Mask Detection system built with OpenCV, Deep Learning and Computer Vision.***
## INSPIRATION
The most effective way of protecting each other during the COVID-19 pandemic is wearing a face mask, thos is where the motivation of creating a system built using deep learning model to identify a person wearing a mask or not, what can be used in crowdy place such as station, schools etc.
## TECHNOLOGY USED
* [OPENCV](https://opencv.org/about/)
* [TensorFlow](https://www.tensorflow.org/)
* [KERAS](https://keras.io/)
* [MobileNet](https://keras.io/api/applications/mobilenet/)
* [Caffe based face detector](https://caffe.berkeleyvision.org/)
## SYSTEM
THE data used in this project is from a Zindi Africa, [The spot mask Challenge](https://zindi.africa/competitions/zindiweekendz-learning-spot-the-mask-challenge) , after participating in the competition I had the idea of using the data to built a face mask detector.
I built a deep learning model `TheSpotmask.ipynb `, transfer learning used pretrained MobileNet application and performed a data augmentation by scaling the images to obtain a very accurate model.
Saved the model and used it to build a system `video.py ` to predict if a person is wearing a face mask or not ina live video stream.
## DATASET
As mentioned earlier, the dataset is from a Zindi Africa competition, download it here [CLICK HERE ](https://zindi.africa/competitions/zindiweekendz-learning-spot-the-mask-challenge/data)
This dataset consists of 3 elements:
* An image zip file that contains the images
* MASK: 1308 images
* NO-MASK: 509 images
* A train labels file
* A sample of the submission file
## RESULT
