The dataset was collected from Kaggle. It comprises pairs of color and depth images (or videos/frames) for face-spoofing detection, categorized into classes: Real (authentic faces) and Fake (spoofed faces). It includes train.csv, validation.csv, and test.csv files. Each CSV contains image_id (file name), with labels included in the training set only. The training images are stored in separate color and depth sub-directories, while validation and test sets contain unlabeled image pairs for prediction. This structured format supports supervised learning, model validation, and final evaluation. All images/video frames are in standard formats, enabling deep-learning pipelines to process multi-modal features for accurate spoof-detection and security-model development.

Part of the dataset is hosted on Kaggle: https://www.kaggle.com/datasets/faber24/lcc-fasd