Intelligent video surveillance for real-time detection of suicide attempts

Wassim Bouachir, Rafik Gouiaa, Bo Li, Rita Noumeir

Abstract: Highlights•A real-time intelligent system for detecting suicide by hanging in prisons is introduced.•A challenging dataset including realistic videos simulating suicidal behavior is created.•An efficient algorithm is used to rebalance the dataset by oversampling the minority class.•Numerous experiments have been done using several classifiers.•A high detection accuracy has been obtained.