An Automatic Illicit Drug Abuser Identification System Based on Facial Features Extracted

Illicit drug abuse is very common among adolescents, and it is a major public health problem that may lead to severe consequences, such as abnormal aging and significant impact on facial appearance, like wrinkles, severe sores, and acne. Long-term substance uses disorder abuse such as methamphetamine, heroin, opioids, and crack can cause complete facial distortion, which makes it a challenging task for many researchers in computer vision applications such as biometrics and visual surveillance. This paper presented an automatic system that identifies illicit drug abusers using a deep convolution neural network (DCCN) to extract deep facia features and selection methods to select only discriminative features that contain rich information. We created new and powerful datasets for normal and illicit drug users from various scenarios. The proposed system outperformed previous works when tested on various experimental scenarios. The proposed system would encourage early interventions to prevent further drug abuse.