The Indian Celebrity Dataset for Face Recognition (ICDFR) is a new dataset compiled from publicly available images of Indian celebrities including Cricketers, Actors, Politicians, Social Workers, Scientists and other Celebrities.

What is the Purpose of ICDFR?

While Labeled Faces in The Wild LFW is good for Face recognition and simple model training, the lack of Asian, Indian faces make it erratic with false positives and trues negatives for predicting Indian faces. We aim to improve the prediction accuracy for Indian faces and make this dataset available to public for free. This dataset will contain only faces of Indian celebrities. The purpose of this dataset is to train faces for better Prediction, Gender, Age and Emotion of Indians.

How is Data collected for ICDFR?

Indian celebrity data is collected from various sources including search, social media, email request, webcam photos. Celebrities have voluntarily provided photos to grow the dataset and support the cause.

How Data is Processed ICDFR?

The photos collected from various sources are analyzed manually for accuracy, quality. Only the celebrity face is cropped from the image and saved to the ICDFR database. Mostly we receive data with group photos of celebrity or with others. We discard other faces for privacy reasons and only the celebrity face is processed.

How is the Data organizedin ICDFR?

The root folder will contain folders of the different categories such as Indian Politicians, Indian Cricketers and inside there will be separate folders for every celebrity (eg. Narendra Damodardas Modi ) with their faces photos.

How can you contribute?

Celebrities or their assistants can send in photos preferably with a clear vision of the celebrity face.

How to Donate?

Donations are welcome and you can donate from Rs.100. All the contributions of 500 and above will be listed in the Donor’s list on site.