2019 IEEE 4th International Conference on Computer and Communication Systems (ICCCS) · 2019 · 48 citations · 18 references
ComputingComputer ScienceComputer VisionMachine LearningPain ManagementDeep LearningPain AssessmentPain CharacterizationPain MeasurementPain PerceptionRepresentation LearningDeep Learning EstimationDeep Learning-based ClassificationDeep Learning-based DiagnosisFacial Pain ClassificationFacial Pain DiagnosisPain Intensity MonitoringPain RecognitionRecurrent Neural Networks
Pain is a primary symptom of diseases and an indicator of a patients' health status. Effective management of pain is important for patient treatment and well-being. There are some traditional self-reported methods for pain assessment, and automatic pain detection systems using facial expressions are developing rapidly; these offer the potential for more efficient, convenient and cost-effective pain management. In this paper, a joint deep neural network model is proposed to classify pain intensity in four categories from facial images. This study used two different Recurrent Neural Networks (RNN), which were pre-trained with Visual Geometric Group Face Convolutional Neural Network (VGGFace CNN) and then joined together as a network to estimate pain intensity levels. The UNBC-McMaster Shoulder Pain database was used to train and test the proposed algorithm. As a contribution to knowledge, this paper provides new information regarding the performance of a hybrid, joint deep learning algorithm for pain multi-classification in facial images.
18
Sinno Jialin Pan, Qiang Yang · IEEE Transactions on Knowledge and Data Engineering · 2009 · 23.2K citations
The “independent components” of natural scenes are edge filters
Anthony J. Bell, T. J. Sejnowski · Vision Research · 1997 · 2.3K citations
Principal components analysis (PCA)
Andrzej Maćkiewicz, Waldemar Ratajczak · Computers & Geosciences · 1993 · 1.6K citations
Painful data: The UNBC-McMaster shoulder pain expression archive database
Patrick Lucey, Jeffrey F. Cohn, Kenneth M. Prkachin et al. · 2011 · 604 citations