Celebrity Face Recognition Project(Overview)

Website UI

• Created a tool that predicts famous celebrities from a scraped dataset images with 78% accuracy.

• Scraped over 1000 images from Google photos using python and selenium.

• Engineered features from the images of every Celebrity to identify face and eyes and cropped the perfect portion by using OpenCV(Haarcascades) and Wavelet.

• Optimized SVM, Random Forest, and Logistic Regression using GridsearchCV to reach the best model.

• Built a client facing UI using JavaScript, HTML and CSS.

• Built a client facing API using flask.

In this Data science and Machine learning project, we classify sports personalities. We restrict classification to only 6 people,

1. Cristiano Ronaldo
2. Maria Sharapova
3. Virat Kohli
4. Roger Federer
5. Sergio Ramos
6. Lionel Messi

Here is the folder structure:

• UI : This contains ui website code
• server: Python flask server
• model: Contains python notebook for model building
• google_image_scrapping: code to scrap google for images
• images_dataset: Dataset used for our model training

Technologies used in this project:

1. Python
2. Numpy and OpenCV for data cleaning
3. Matplotlib & Seaborn for data visualization
4. Sklearn for model building
5. Jupyter notebook, visual studio code and pycharm as IDE
6. Python flask for http server
7. HTML/CSS/Javascript for UI

Sources

Image Scrapper Code: https://towardsdatascience.com/image-scraping-with-python-a96feda8af2d

Image Scrapper Code: https://medium.com/@wwwanandsuresh/web-scraping-images-from-google-9084545808a2

You can view on the details of this project here: https://www.youtube.com/channel/UCh9nVJoWXmFb7sLApWGcLPQ