CED-Net: context-aware ear detection network for unconstrained images

Personal authentication systems based on biometric have seen a strong demand mainly due to the increasing concern in various privacy and security applications. Although the use of each biometric trait is problem dependent, the human ear has been found to have enough discriminating characteristics to allow its use as a strong biometric measure. To locate an ear in a face image is a strenuous task, numerous existing approaches have achieved significant performance, but the majority of studies are based on the constrained environment. However, ear biometrics possess a great level of difficulties in the unconstrained environment, where pose, scale, occlusion, illuminations, background clutter, etc., vary to a great extent. To address the problem of ear detection in the wild, we have proposed two high-performance ear detection models: CED-Net-1 and CED-Net-2, which are fundamentally based on deep convolutional neural networks and primarily use contextual information to detect ear in the unconstrained environment. To compare the performance of proposed models, we have implemented state-of-the-art deep learning models, viz. FRCNN (faster region convolutional neural network) and SSD (single shot multibox detector) for ear detection task. To test the model’s generalization, these are evaluated on six different benchmark datasets, viz. IITD, IITK, USTB-DB3, UND-E, UND-J2 and UBEAR, and each one of the databases has different challenging images. The models are compared based on performance measure parameters such as IOU (intersection over union), accuracy, precision, recall and F1-score. It is observed that our proposed models CED-Net-1 and CED-Net-2 outperformed the FRCNN and SSD at higher values of IOUs. An accuracy of 99% is achieved at IOU 0.5 on majority of the databases. This performance signifies the importance and effectiveness of the models and indicates that the models are resilient to environmental conditions.