18 December 2017 in IEEE Access
Iris recognition refers to the automated process of recognizing individuals based on their iris patterns. The seemingly stochastic nature of the iris stroma makes it a distinctive cue for biometric recognition. The textural nuances of an individual's iris pattern can be effectively extracted and encoded by projecting them onto Gabor wavelets and transforming the ensuing phasor response into a binary code - a technique pioneered by Daugman. This textural descriptor has been observed to be a robust feature descriptor with very low false match rates and low computational complexity. However, recent advancements in deep learning and computer vision indicate that generic descriptors extracted using convolutional neural networks (CNNs) are able to represent complex image characteristics. Given the superior performance of CNNs on the ImageNet large scale visual recognition challenge and a large number of other computer vision tasks, in this paper, we explore the performance of state-of-the-art pre-trained CNNs on iris recognition. We show that the off-the-shelf CNN features, while originally trained for classifying generic objects, are also extremely good at representing iris images, effectively extracting discriminative visual features and achieving promising recognition results on two iris datasets: ND-CrossSensor-2013 and CASIA-IrisThousand. We also discuss the challenges and future research directions in leveraging deep learning methods for the problem of iris recognition.
4 February 2021 in IEEE Sensors Journal
Epilepsy is one of the most prevalent neurological diseases among humans and can lead to severe brain injuries, strokes, and brain tumors. Early detection of seizures can help to mitigate injuries, and can be used to aid the treatment of patients with epilepsy. The purpose of a seizure prediction system is to successfully identify the preictal brain stage, which occurs before a seizure event. Patient-independent seizure prediction models are designed to offer accurate performance across multiple subjects within a dataset, and have been identified as a real-world solution to the seizure prediction problem. However, little attention has been given for designing such models to adapt to the high inter-subject variability in EEG data. We propose two patient-independent deep learning architectures with different learning strategies that can learn a global function utilizing data from multiple subjects. Proposed models achieve state-of-the-art performance for seizure prediction on the CHB-MIT-EEG dataset, demonstrating 88.81% and 91.54% accuracy respectively. In conclusion, the Siamese model trained on the proposed learning strategy is able to learn patterns related to patient variations in data while predicting seizures. Our models show superior performance for patient-independent seizure prediction, and the same architecture can be used as a patient-specific classifier after model adaptation. We are the first study that employs model interpretation to understand classifier behavior for the task for seizure prediction, and we also show that the MFCC feature map utilized by our models contains predictive biomarkers related to interictal and pre-ictal brain states.
1 December 2017 in Pattern Recognition
21 June 2018 in IEEE Robotics and Automation Letters
Accurate localization of crop remains highly challenging in unstructured environments, such as farms. Many developed systems still rely on the use of hand selected features for crop identification and often neglect the estimation of crop quantity and ripeness, which is a key to assigning labor during farming processes. To alleviate these limitations, we present a robotic vision system that can accurately estimate the quantity and ripeness of sweet pepper (Capsicum annuum L), a key horticultural crop. This system consists of three parts: detection, ripeness estimation, and tracking. Efficient detection is achieved using the FasterRCNN (FRCNN) framework. Ripeness is then estimated in the same framework by learning a parallel layer which we experimentally show results in superior performance than treating ripeness as extra classes in the traditional FRCNN framework. Evaluation of these two techniques outlines the improved performance of the parallel layer, where we achieve an F 1 score of 77.3 for the parallel technique yet only 72.5 for the best scoring (red) of the multiclass implementation. To track the crop, we present a vision only tracking via detection approach, which uses the FRCNN with parallel layers as input. Being a vision only solution, this approach is cheap to implement as it only requires a camera and in experiments, using field data, we show that our proposed system can accurately estimate the number of sweet pepper present, to within 4.1% of the visual ground truth.