LNS Occluded and Low-Light Human Face Detection Dataset
Description
The Hybrid Face Detection Dataset: Occluded and Low-Light Facial Images is a comprehensive and diverse collection designed to support the development and evaluation of robust face detection algorithms under real-world challenges. This dataset contains a total of 12,000 images, primarily featuring human faces captured under a variety of difficult conditions. It is an enhanced and updated version of the LNSONI Human Face Dataset, with expanded coverage of complex scenarios and demographic diversity. Key features of the dataset include: Low-Light Face Images: A significant portion of the dataset includes images captured in dim lighting or with poor illumination, simulating real-world conditions such as night-time surveillance, shadowed environments, or indoor low-light scenes. Occluded Face Images: The dataset contains faces partially covered by common occluding objects such as masks, sunglasses, scarves, hands, and hair. These examples are essential for training models to be resilient against partial visibility. Multi-Angle Face Images: To increase variability, the dataset includes images of faces from various angles and poses, including frontal, profile, and semi-profile views. This helps in evaluating pose-invariant detection capabilities. Non-Face Images: To improve classifier discrimination, the dataset also includes a curated selection of non-face images featuring background objects, textures, and clutter that could potentially trigger false positives in detection algorithms. Diversity of Subjects: The dataset spans a wide range of age groups, from infants to elderly individuals, and represents multiple ethnicities and geographical regions, ensuring inclusivity and aiding generalization across global populations. This dataset is suitable for training, validating, and benchmarking face detection systems, especially those designed to handle occlusion, low illumination, and variability in face orientation. It supports research in areas such as smart surveillance, biometric systems, mobile authentication, and real-time face recognition in challenging environments.