I failed at a Hackathon. Did I?

Our yardstick for success is sometimes unreasonable.

How do you feel when you’ve spent hours on something, only to see it end with a fizzle? That’s how I felt last Friday at my first hackathon at Gramener. The theme was ‘movies’ and the datasets were so exciting that I had even tweeted about my participation.

The plan was to apply a facial recognition exercise on Bollywood movie posters & determine the facial emotion of characters on each poster. My mouth was watering thinking of the results of the analysis — a folder full of posters with ‘Happy’ characters, another folder with characters looking ‘scared’ and so on. Did I get what I want? I did get a result, but the quality was so poor I wanted to bang my head against a wall. The hard part was that I had spent 2 days and an entire night just to scrape the posters & install the libraries. Yes, I know it was only a day long hackathon, but who knew it was going to take a lifetime to just install the libraries. After all that effort, to end up with a clumsy result was agonizing.

Feeling defeated and small, I regretted for having ventured into something like this. Felt like I should have just shut up and binge watched some Netflix series or something. After two days of feeling miserable about this ‘failure’, I’ve realized that I’ve been too hard on myself. Sure the disappointment of not being able to create a hackathon output exists, but when I think of how much I learnt during the creation process, it surprises me.

So here it goes;

Great! So now I can install the face recognition library with ease? Nope.

What followed next was an all-nighter trying to get the below to work;

But fate had other plans. The model did a decent job only on frontal faces and my output went horribly wrong. The folder which had to contain posters that had characters exhibiting an angry emotion had the following output. Crazy, isn’t it?

By now it was 5:00 am & I could hear Bandra’s early morning walkers gearing up for a beautiful day. Tired, disappointed & close to tears I crawled to bed.

I had failed. That was it. I wasn’t good enough. But now, two days later, I’ve realized that it was not a disaster after all. At least I got Anaconda 3 & the face recognition library installed. I also have realized that the library would give better results with frontal faces. So, you know what I’ve planned next? Yikes! I’m not going to risk a hastily made announcement again. I’ll keep you posted when it is done. Oh, and by the way, please remember — It’s not just about the destination, the journey matters.

Related posts:

If my unique experiences can be translated from my microcosm to a wider world view for you in order to draw some universal truths from them, so be it… I spent 16 years as a professional touring…

When I walked into the bedroom and saw her form sleeping in the bed, I was immediately filled with lust. The day was yet to full begin and I had fresh morning energy coursing through me. Her dark…

To study the distribution of sexist attitudes from a sociological perspective, we differentiate between characteristics from the parents ( ascribed social characteristics) and young people’s own…