An Ensemble of Deep Convolutional Neural Networks for Drunkenness Detection Using Thermal Infrared Facial Imagery

Concurrent Convolutional Neural Networks with Decision Fusion to Diagnose COVID-19 using Chest X-ray Imagery

“…The model proposed in this paper is based on a competitive system built of two identical convolutional neural networks with decision fusion. The idea of the concurrent neural networks has been inspired by the models of Neagoe et al described in [12], [13] and [14]. In [12] Neagoe et al have proposed a new type of neural classifier, composed by a lot of concurrent self-organizing maps (CSOM); this has been applied with very good results for facial image recognition and for multispectral pixel classification.…”

“…In [13] CSOM model has been successfully used for change detection in satellite imagery. In [14] Neagoe et al proposed an ensemble of two CNNs with a small architecture asymmetry using decision fusion for drunkenness diagnosis in thermal imagery. Our proposed model is composed of two identical CNNs.…”

Modification Method for Facial Skin Temperature Image Considering Facial Artery Structure

“…Based on this information, many studies have estimated various human states such as stress, emotions, and drowsiness. These studies extract features for estimation by defining arbitrary rectangular regions [1] or by applying machine learning methods such as a Convolutional neural network [2]. However, given that the information on the blood flow is used for estimation, modifying the thermal face image considering facial artery structure before feature extraction could be useful in obtaining more valuable information from the thermal face image.…”