Me, Myself and My Killfie: Characterizing and Preventing Selfie Deaths

Authors: Hemank Lamba, Varun Bharadhwaj, Mayank Vachher, Divyansh Agarwal, Megha Arora, Ponnurangam Kumaraguru

Our world is becoming smaller with time, bringing us closer and bestowing upon us a number of avenues to easily showcase ourselves in any manner we want. Perhaps the biggest facilitating agent in this regard, is Online Social Media (OSM). In a way, OSM replicates our world, with friends, interactions and constant information exchange. The world of OSM seems to have developed an interesting currency of its own too – LIKES and COMMENTS, the dollars and cents of the virtual realm; something which everyone aspires to have in abundance.

We are also familiar with the popular “selfie” phenomenon. Recognized as the “word of the year” by Oxford dictionary in 2013, the “selfie” is defined as a “photograph taken of oneself, and uploaded to a social media website.”  In recent years, there has been a sharp increase in the number of selfies posted on OSM. However, one particularly disturbing trend that has emerged lately is that of clicking dangerous selfies; proving to be so disastrous that during the year 2015 alone, there have been more deaths caused due to selfies than shark attacks all over the world [1]. Figure 1 shows examples of such selfies taken moments before the fatal incident. A selfie-related death can be defined as a death of an individual or group of people that could have been avoided had the individual(s) not been taking a selfie.

The level of threat that adventurous selfie taking behaviour exposes people to, is being acknowledged slowly by governments as well. Russian authorities came up with a public awareness campaign to enlighten citizens of the hazardous implications of taking selfies [2]. Similarly, Mumbai police recently classified 16 zones across the city as No-Selfie zones, after a rise in the number of selfie casualties [3].

The reason for this outrageous trend of dangerous selfies becomes clear when we combine the thoughts above. Since the advent of online social networks, people have developed an insatiable urge to be the most “popular” in their community. In medicinal terms, this has been long compared to forms of narcissism and in relation to selfies, termed as Selfitis [4,5,6]. This becomes the prime reason why people resort to performing risky feats while taking a selfie to garner more appreciation in the form of likes and comments from their friends online.

We, at Precog@IIITD chose to analyse the issue from a technical perspective and to dive deeper into what characterizes a selfie casualty/death, what kind of information we can extract from selfie images and how selfie casualties can be prevented.

Over the past two years, we found that a total of 127 deaths have been reported to be caused due to selfies, of which a whopping 76 deaths occurred in India alone! [7] Table 1 shows the country-wise distribution of selfie casualties across the world. The reasons for these selfie casualties were found to broadly belong to the following categories (Figure 2) at

  • Height Related – Selfie casualties caused due to people falling from an elevated location. [8]

  • Water Related – Selfie casualties caused due to drowning. [9]

  • Height and Water Related – Selfie casualties involving falling from elevated locations into a water body. [10]

  • Vehicle/Road Related– Selfie casualties caused due to vehicle accidents. [11]

  • Train Related– Selfie casualties caused due to being hit by a train.[12]

  • Weapons Related– Selfie casualties caused due to accidental firing of a weapon.[13]

  • Animal Related– Selfie casualties caused due to attack by an animal while taking the selfie with or near the animal.[14]

  • Electricity Related- Selfie casualties caused due to electrocution from live wires.[15]

Figure 2: (a) Number of Deaths and (b) Number of Incidents due to various reasons

Using a collective dataset of 138,496 tweets collected between August and September 2016, we implemented a three-fold architecture based on Image features, Location features, and Text features to quantify the danger level of selfies in our dataset.  Our machine learning model takes into account a variety of features to identify dangerous selfies along with their potential risks, and analyses common characteristics in these images. These features are supplied to four different classifiers with similar parameters to avoid bias in the results. Table 2 shows the sets of features we used for each feature type.

Table 2: Location-Based, Image-Based and Text-Based features used for classification of selfies

After thorough analysis, we found that the image-based features are the best indicators that accurately capture the dangerous nature of a selfie, in comparison to other feature-types. This seems logical as image features attempt to infer meaning directly out of the image, in a sense replicating our visual senses. Our model resulted in an accuracy of 73.6% for the task of identifying a dangerous selfie.

To further capture the risk type of a dangerous selfie, we used specific features that were relevant only to a particular risk type and supplied the data to our classifier. In particular, we concentrated on singling out dangerous selfies that belonged to height, water and vehicle related risks. We found that the set of features performing the best for this task was a combination of all 3 feature types – Image, Location and Text based features, and the best accuracy was obtained on the Water-related features. With remarkable accuracy, we have been able to establish a method to identify and capture the “danger level” of a selfie along with its risk type.

With the growing trend of dangerous selfies, it becomes important to spread awareness of the inherent hazards associated with people risking their lives simply for the sake of recognition on a virtual forum. As Shakespeare coins it, this type of “Bubble Reputation” induced by a dangerous selfie posted on OSM has claimed multiple lives lately. This work is a small contribution towards making the world safer, by making the people aware.

Our full report / paper on this work. You can access the portal and our dataset here.





[4] S. Bhogesha, J. R. John, and S. Tripathy. Death in a flash: selfie and the lack of self-awareness. Journal of Travel Medicine, 23(4):taw033, 2016

[5] B. Subrahmanyam, K. S. Rao, R. Sivakumar, and G. C. Sekhar. Selfie related deaths perils of newer technologies. Narayana Medical Journal, 5(1):52–56, 2016.

[6] A. LAKSHMI. The selfie culture: Narcissism or counter hegemony? Journal of Communication and media Studies (JCMS), 5:2278–4942, 2015










Precog: The Phenomenon

A normal afternoon day, with the heat scorching its way through to every one of our rooms at NIT Trichy. I was scrolling through my mails, as usual. However, today something caught my eye. A particular mail from my ACM student membership subscription that looked like this:

I am a huge fan of Social Networks research by the way. Right from its theoretical aspects – centrality measures, the small world phenomenon, etc.. up to its more applied research aspects – mining important data from Online Social Media, creating core systems that constantly evolve and adapt using the humongous data obtained from social networking sites, OSM Analysis has never failed to amuse me.  So the moment I read and researched about Dr.PK, I knew he was one professor anyone would really desire to work with, and that this is the kind of work I would like to pursue.

From looking through the window, To being welcomed inside

The most significant characteristic about this group is that, Precog means business. They are a bunch of researchers who have made their mark all over the world simply by doing what they love, bounded by no restrictions whatsoever.  This is what made this research group an even more desirable organization to be a part of. The urge to be a part of this only increased every day ever since I read about it. I applied to PK soon after, and after a structured and elaborate induction process, on another sunny afternoon, a mail popped up in my inbox.

I got my offer letter.

“Make us proud, by making yourself proud.”

 PK’s words from his mail to me lingered in my mind as I sat in the CERC lounge on my first day, waiting for my mentor and PhD scholar Prateek Dewan to brief me on my project. Prateek came, we exchanged pleasantries and got down to business immediately. The words Prateek said:

“You’ll be working on the Image Analysis domain for this summer. In particular, you will be working on Extracting the Sentiment (Emotion) from an Image. All the Best.  Always here to help you if you need anything.”

The meeting lasted 10 minutes. I was taken by complete surprise. The most image analysis I had ever done before was to see if an image on my Facebook news feed was worthy of a like or not. 😛

But that’s the very beauty of Precog. Remember how I told you about there being no restrictions whatsoever here? This was a standing example of that. My objectives were clear. But I had COMPLETE freedom to pursue any path I wanted to get the job done by the end of the summer. [1] Thus said, I embarked on a journey of pure knowledge discovery and fun.

 The Liam Neesons of Precog

Hardly a few weeks had gone by during my internship, and I was LOVING it here. My work was going on a smooth pace. I was slowly getting a hang of the state-of-the-art in Image Analysis and constantly surveyed relevant literature. Of course, I was nowhere near even charting a course to achieve my goals for the summer, but I knew I was on my way. I had lots of motivation and initiative, but at regular intervals of time, I also needed direction, perspective and feedback. Here is where I introduce the Liam Neesons of Precog, or as PK calls them, the “Pillairs” of Precog. Why Liam Neeson though? Well, that’s because every member at Precog has a “very special set of skills” that make them really powerful and strong in their domain of computer science. Each member here specializes in something unique, working on a completely different problem altogether but with just one common cause – to help the common man and do social good. Needless to say, this eclectic mix turned out to be a gold mine of information for me. Every interaction I had with each one of them, there was some take away from it. The weekly brainstorming sessions we had, discussing important and high-impact papers were a huge repository of information too.

“In Precog, data and information were not just something we worked on, but something we gained too. “ [2]

But none of this would be complete without mentioning the pioneer behind it all, PK. PK always kept us on our toes by pointing us to interesting articles/research/technology as it released and encouraged us to probe more into each one of them. Perhaps the one person with the highest motivation levels in Precog is PK himself. There was never a dearth of motivation in the lab. PK always made it a point to keep throwing new ideas in the air, and if something seemed viable, included everyone in bringing up a solution. He had this knack of just sparking a small thought in our minds and leaving it at that. That thought would transform into an idea, the idea into something else until it lead to some really big solution for a confounding problem. His heartfelt and personalized greetings, wishes and guidance will put you in utter disbelief – how could such a busy and accomplished person still have time for each one of us and care about each of us and our projects with such attention? All of this made me grow very attached to PK and the group as a whole, to the extent of worrying about my fast approaching last day. 

The Technical Details

Keeping this part very short, with constant guidance from my mentor, we were able to propose two novel architectures – one to find the expression of all faces present in an image using special facial features and the second to classify an image as a whole as positive or negative based on a deep network architecture. Never had I thought I would be successful in establishing a working model for this if not for the constant support of PK, my mentors and all members of Precog as a whole.

 The Fun Parts!!

Apart from all the rigorous working, pulling all-nighters and meeting deadlines, the fun quotient was also extremely high here at Precog. We went on a lot of outings, right from nearby ice-cream desert shops to malls in Noida. Recounting all those experiences now, as an intern, I felt like I experienced almost everything there is to experience at Delhi. How much ever Precog works, they have fun equally or even much more! Ultimately, be it a tough problem to crack, or a big pizza to finish, we did it as a group, together. 😛

 We go live in 5…4….3…

As a result of all the collective hardwork of the Image Analysis team I was a part of, Helix, an Image Analysis tool was born. I now reminisce over my internship and think about how the journey was as wonderful as the elated feeling I get when I see a small part of my work being used by live users everyday. The fun discussions I had with the PhD scholars, the homely feeling I got every time I entered the lab, all the fun I had with my newly found friends, and of course all the knowledge I gained – everything was worth it at the end. What I was when I started, and what I am now.  It was then that I knew, this was the true goal of my internship. [3]

With a sense of fulfilment and content, my internship was complete. Not a day goes by, when I don’t see the Precog sticker on my laptop and recount the wonderful experience I had at Precog.

A pic of me with my mentor and close friend, Prateek:

The Head Fake

No Precog blog is complete without Randy Pausch’s wisdom being in it. As Randy describes it, A head fake is something that we are made to do, but its purpose is completely different from what is presumed. Observe above in the blog that I have put the numbers [1], [2] and [3] at several points. Those were the head fakes of my internship.

[1] This internship of mine was not solely aimed at the result alone. The freedom I got to pursue my desired path throughout the internship lead to a huge inflow of ideas and thoughts in my mind. This refined my thinking, not just for the problem, but for everything else too. I wasn’t fed fish. I now knew how to fish. This was my first head fake.

“I knew how to fish.”

[2] By the end of the summer, I had more skills than I could possibly have expected to get just by working on my problem alone. This is completely attributed to the sessions we had, information we shared and discussed.  I had so much more skills now that I could confidently add to my kitty than I could imagine. This was my second head fake.

“I too got a special set of skills.”

[3] Finally, I got a family. A permanent set of close friends, and an evergreen connection to Precog, no matter where I proceed further in life. This was my third head fake.

“Once a Precog-er, always a Precog-er.” 

For all the prospective future students reading this blog:

The final head fake: The blog was not just to pen down my experience at Precog. It was to motivate you guys to come and experience the phenomenon that is Precog. 🙂

If you are interested in #PrecogSummer 2017, please do checkout