Preventing Sexual Harassment in India Using AI and Heatmaps

Background

In India, sexual harassment is a pervasive issue, with UN Women noting that 1 in 3 women globally face some form of sexual assault in their lifetime. In India, a rape occurs every 20 minutes. Cultural barriers, fear of victim-blaming, and distrust of law enforcement prevent many women from reporting these incidents. Safecity aims to empower victims and has built one of the largest global datasets on sexual harassment cases.

Objective

The project aimed to utilize Artificial Intelligence to develop solutions that increase the safety of women in major Indian cities by creating tools that predict and help avoid areas with high risks of sexual harassment.

Approach

The Omdena community collaborated with Safecity to build solutions using open-source data and Safecity’s extensive database. Key technologies included:

• Heatmaps: Visual representations of harassment hotspots.
• Safe Routes: Calculated using Nearby Search, Directions API, and Grid Coverage techniques.
• Machine Learning: Developed a sexual harassment category classifier with 93% accuracy using convolutional neural networks and LSTM models.

Results and Impact

The collaborative effort resulted in AI-driven tools that accurately identify high-risk areas and suggest safer travel routes for women. These solutions have been integrated into the Safecity platform, which crowdsources personal stories of harassment to further refine data accuracy. The project not only increased efficiency but also empowered women by providing reliable tools for safety planning.

Future Implications

The success of these AI-powered solutions paves the way for future policy changes, encourages the development of similar tools in other regions, and promotes further research into AI applications for public safety. By demonstrating the effectiveness of data-driven approaches, this project could influence how governments and organizations address sexual harassment on a broader scale.