The Intra-MIC Hack: A Word from the Teams (Part 3)
We at SRMMIC had conducted an intra-MIC hackathon from 5th September to 18th September as an effort to inculcate a sense of belonging among the new members. The motivation was to encourage everyone, old and new, to participate and interact with people and come up with a new hack in 2 weeks! There were no specific themes, everything machine intelligence and beyond was fair game. A total of 12 teams participated, and the four teams for this blog are: Sodium, Magnesium, Aluminum and Silicon
Team Sodium
Team members: Indira Dutta, Atharv Shah
What was your hack about?
Our project was a binary classification problem that dealt with predicting if a customer of a popular Asian Music Streaming site will continue churning after a specific date. All the famous video/audio streaming sites these days want to retain their customers in order to prevent their sales from decreasing. A detailed report of their customers would help them to meet the demands and hence enhance their performance.
What did you learn along the way?
We learned how to work on massive large datasets. Another major area was learning about a lot of new algorithms that we had never worked previously on. We also learned how to work as a team and how to learn from our mistakes.
What difficulties did you face?
Dataset preparation was one such important challenge we faced because we needed to select the most important features that would give us the best results. Hyperparameter tuning for the gradient boosting algorithms we used took a lot of time to execute as well. Finally manual feature engineering was incredibly complex as well.
Any special comments?
It was quite the unique experience! Looking forward to season 2 of the hack!
Check out the hack here!
Team Magnesium
Team members: Prakhar Singh, Aakriti Kinra, Devansh Pratap Singh
What was your hack about?
Being beginners, me and my teammates thought of going with something simple, yet informative. So we decided to go with “Handwritten Digits Recognition” using Python, Tensorflow, Keras and OpenCV. We trained a simple FFNN model which recognized human handwritten digits from the input, a scanned image of a handwritten number. The output was composed of the number entered along with the feature map. We used the MNIST dataset that contains 60,00 training images and 10,000 test images. We plan to expand the project to recognize full sentences.
What did you learn along the way?
We had no prior experience of working with Tensorflow, Keras, Matplotlib and OpenCV, so we decided to take up an online course. Along the way, we worked on multiple small programs to understand the concepts. We were able to understand concepts behind CNNs and wish to apply these in future projects. Apart from technical stuff, it was our first time working in a team, so we would like to mention teamwork and time management as well.
What difficulties did you face?
We completed the hack smoothly without any problems as such except for TensorFlow which was a bit tough. Besides, we had errors due to very small typos which we were unable to debug easily. One of our teammates was sick throughout the hack so the workload was much and we had to strategize all over again. Since we were learning it all for the first time and applying them at the same time, the allotted time was a bit less and we had to follow a hectic schedule.
Any special comments?
This hack helped us solidify our learnings as well as participate in a team of equals. We just had great fun overall!
Check out the hack here!
Team Aluminium
Team members: Yudhajeet Bhattacharya, Annanya Pandey, Rishav Banerjee
What was your hack about?
Our hack was a letter recognition for Hindi. We created a web app with a canvas where the user could draw a letter in Hindi and they would get the name of the letter. We thought it’d be a unique project where we’d get to learn and implement new skills.
What did you learn along the way?
As beginners and intermediates to the field, we learnt a ton about image processing and the dl framework keras. There was also streamlit, so we learnt about deployment as well.
What difficulties did you face?
We faced a little difficulty with regards to the deployment of the model as well as getting the accuracies right. But it all came out pretty decently at the end!
Any special comments?
It truly was a great opportunity where we learnt a lot. Looking forward to similar activities!
Check out the hack here!
Team Silicon
Team members: Swarnabha Das, ANV Sreevishnu, Palveet Kaur Saluja
What was your hack about?
Covid-19 has made such a big impact on the world that going to a hospital for consultation has become scary. Poor accessibility to healthcare across India has made it worse. Our vision was to digitize healthcare in India where any person with a smartphone can have access to healthcare. Our web app helps in smooth and convenient communication with the doctor besides providing features like self assessment and inkblot test which would help a person keep a track of their mental health, a chatbot for helping the users, a videocall option in order to consult the doctor as soon as possible and a feature to access nearby hospitals and pharmacies.
What did you learn along the way?
In the course of time, we learnt to use OpenCV to build a simple face recognition model. We learnt to recognise faces not just through pictures but from a live video as well. Managing so many GET and POST requests using Flask was new to us not to mention adding databases and live streaming webcam in Localhost . Besides technical stuff, we also learnt a lot about psychological medical tests that can be done sitting online.
What difficulties did you face?
The very basic thing to keep in mind was designing the UI as sober as possible as it is a medical based app which was eventually successful. Integrating webcam(cv2) with flask and recognising face from the live video through webcam was quite difficult. Connecting database and retrieving data as per user request and integrating Chatbot in the web also posed some difficulties.
Any special comments?
We would like to convey our special thanks to the entire organizing team for pulling off the MIC Hack smoothly. Hope to have many more such hacks in the future and be a part of them.
Check out the hack here!
