Using AI for women’s safety

3 min readOct 5, 2019

Safecity collaborated with OMDENA on an AI challenge. Here’s what we learned and why is it important.

This past week I spoke with a group of students at the University of District of Columbia about crowd mapping sexual violence in public spaces. . I explained to them how the anonymous stories women and girls share on my organization Safecity’s website gives us clues on location based trends and patterns that contribute to the violence.

One of the students mentioned that some of the DC metro subway stations attract predators around when they are shutting down at night. In fact, she recounted an incident when she was attacked at the metro at Brookland Catholic University. She mentioned that the station was poorly lit and she did not see her attacker till it was too late. In a separate conversation with a male technologist who works with Open Street Maps in DC, he confirmed that the subway at Brookland Catholic University is dimly lit and even to him, it feels a bit awkward walking into a space that felt like a cave.

What struck me is how dark spaces can become the “comfort zone” of a perpetrator and it certainly seemed to be the case at this station. The ability to make these correlations and find common patterns and trends that could indicate problematic locations is fascinating. Could we understand better what contributes to the violence and is there a way to allocate resources more effectively for prevention? This is what the women’s safety audit created by women in Toronto around 30 years ago does. The UN and other agencies use it all over the world to evaluate factors like lighting and graffiti and correlate it with how safe people feel.

That’s what my organisation aims to do with our crowd map Safecity with personal stories of women and girls. We started this initiative about seven years ago as an immediate response to a horrific gang rape of Jyoti Singh on a bus in Delhi. Since then we have collated over 12000 personal stories from various parts of the world but mainly India and Kenya. These stories which are largely anonymous, form a new dataset which does not exist in official records due to the under reporting of sexual violence.

Over the past few months, we partnered with OMDENA, a global platform which works to build Artificial Intelligence models for social good via harnessing the power of a global collaborative community of data scientists, technologists and engineers. There were 32 volunteers working remotely from 14 countries spanning five continents.

It was an interesting exercise in predictive modelling where the Safecity dataset for Mumbai and Delhi were used to identify probability factors of “safe” and “unsafe” spaces. Various techniques of modelling were used to layer data including infrastructure like schools, colleges, hospitals, cinema theatres, public parks and surrounding areas to get a sense of what risk factors might be involved. The diversity of the participants added to the various perspectives of “safe spaces” and potential to identify “safe routes”.

Analysing data through the stories was a powerful way to understand the issue of sexual violence which is a global pandemic. According to UN Women, one in three women around the world experience some form of sexual assault at least once in their lifetime. Yet there is severe under-reporting making the problem “invisible”. This crowdsourced data set helps make the issue more visible, gives tangible data points and helps people understand the nuances of the problem better.

As with any social issue, the lack of data and thereby the lack of understanding prevents effective deployment of resources even if there is a mandate to solve it. So, with such exercises like hackathons and predictive models, we can try to fill the gaps in understanding and also show what solutions can be possible. The goal of these efforts is prevention rather than punitive measures.

The passion, time, skill and energy invested by people from different parts of the world during the OMDENA challenge was exciting and motivating. We believe we have allies who are determined to help us find solutions and end violence against women and girls. We also believe that through these stories, the participants will be better aware of the issue and active bystanders in their daily life. We are grateful to OMDENA for partnering with us.




Contradiction of sorts — dreamer and doer, introvert yet extrovert, grounded whilst always flying. Feminist. Know more;