Domestic Violence Detection System Using Natural Language Processing
Fathima K S, Rosna P Haroon, Fitha Fathima, Thameesa Thajudeen, MuhammedShera Fudin · 2023
Domestic violence is a widespread problem that leaves victims feeling helpless and trapped. It refers to a pattern of abusive behavior in an intimate relationship, where one partner seeks to exert power and control over the other. To address this issue, we proposed a system that incorporates several advanced technological aspects to address domestic violence and support victims. It includes a speech recognition module that identifies abusive behavior patterns through specific words and phrases associated with domestic violence. Natural language processing techniques analyze the context and tone of the language to understand the emotional state of the victim and detect instances of emotional abuse. A camera system with computer vision capabilities detects physical violence and abuse signs through object detection and tracking in video footage. The system is accessible through a mobile app, enabling victims to collect evidence of abuse and access necessary resources and support services. Machine learning algorithms continuously improve the system's accuracy and performance by learning from a large dataset of abusive language patterns and visual cues. When abusive behavior is detected, the system can automatically send alerts to law enforcement and emergency contacts, providing immediate assistance. Recorded voice and video footage, along with associated analysis results, are securely stored as evidence for legal proceedings. By integrating these technological aspects, the system aims to empower victims, improve evidence collection, and support efforts to combat domestic violence.