This repository contains a data collection of over 11000 pedestrian images taken in the wild. The data was collected from a variety of sources in the hope of creating a benchmark dataset for pedestrian face detection / recognition research. The dataset contains over 55000 detected faces. The dataset is freely available for academic research purposes.

Download the Real Face dataset through Google Drive, unzip the downloaded file and you will get the following structure:

.
├── 1.jpg                   # Image file
├── 1.xml                   # Annotations file
├── ...
├── 11437.jpg
└── 11437.xml

<annotation>
	<object>
		<name>face</name>
		<pose>Unspecified</pose>
		<truncated>0</truncated>
		<difficult>0</difficult>
		<bndbox>
			<xmin>773</xmin>
			<ymin>437</ymin>
			<xmax>796</xmax>
			<ymax>468</ymax>
		</bndbox>
	</object>
	...
</annotation>

For questions and result submission, please contact Leonardo Ramos at [email protected]