Mind, can either be our best friend or the worst enemy. Mental health and mental illness are often considered to be taboo, something society doesn't like mentioning. But what I have been through is worse than society's judgement could ever be. I am also afraid of being looked down upon for going through something that is completely normal. So normal that 1/4 adult suffers from some sort of mental illness.

There are 38,000 suicides each year in the US. That means one about every 15 minutes. And 90% of suicides are related to a mental illness, depression bipolar disorder, schizophrenia, anorexia, borderline personality and the list goes on!

Our attitude towards this problem needs to be changed!

What it does

AI-based Emotion and facial expression detection

Psych-o-tech tremendously increases efficiency and access to medical attention. Improving the connectivity of mass and hence helping numerous mental illnesses in significantly less time! Hence, the cost of therapy reduces, the quality of treatment remains unchanged, so many people learn to smile again and so many doctors can give their undivided attention to the patient knowing that Psych-o-tech has got their back. WIN-WIN

How we built it

We divided the challenge into two vast components:

  • Remote Psychotherapy Sessions: remotely diagnosing an illness via face-to-tace conference. Psych-o-tech will help in understanding the emotional behaviour shown by the patient. The dataset we are using has been taken from Kaggle. The link must be mentioned in the references section. The patient's record has been plotted in the form of a graph next to the meeting division, which ensures real-time emotion detection. This helps therapists to focus more on patient's therapy and less on note-taking. This in turn makes sure that the meetings are extremely efficient and records are organised properly.
  • Local Raspberry Pi Sessions: These are traditional therapies, but this time, therapists have an assistant to help them with maintaining records of patient's emotional status. This is very similar to implement, just this time we use Raspberry Pi to do machine learning.

Challenges we ran into

Before the hackathon, none of us was proficient in any particular domain. The idea sounded crazy and difficult to inculcate. Embedding live-video calls, graphically representing emotions and

Accomplishments that we're proud of

We developed a website, that we learnt from AngelHacks workshop itself. We self-taught ourselves basic Javascript and used it in site development. No one had experience with AI or Machine Learning, we thought it would be easy to use DeepAI, and we really wanted to try that out, but then we researched a bit and found a dataset from Kaggle. It was a new and exciting experience to practically learn and implement these models in real life and TRY to solve such an important social issue.

What we learned

What we have made in here, is mostly what we have learned from here itself! Flask framework was really difficult for us to use as none of us had any experience with the same. We learnt it in the stipulated amount of time and used it.

What's next for Psych-o-tech

We are still working on improving our web-page aesthetically, our AI model intellectually. It has been kept open source so it can reach out to the maximum people possible. Everyone is encouraged to contribute to the repository.


Should we simply sit here and pretend like we shouldn't talk about mental illness because talking about it is weird, those people are crazy. Let's get it straight that one in four that have a mental illness versus the three in four, these three in four are those crazy ones.

We can have a very human conversation about a very human experience, and it won't be hard. Just be HUMAN and open your heart because you know, if it's not your friend, it could be you.


Kaggle Dataset (to be updated)

Share this project: