Auto Segmentation of Sub-Retinal Fluid for Central Serous Retinopathy Pathology
This is my year long major project for my Bachelor's degree. We used Capsule Networks to improve the auto image segmentation performance of the prior approaches.
We developed a new architecture based on the capsules concept, producing comparable performance to existing state of the art approaches but with
increased computational efficiency. We hope to publish our findings soon.
Generative Adversarial Networks (GAN)
I understood the basic concepts of GANs and implemented a vanilla Softmax GAN for the MNIST dataset and a Triple GAN for the CIFAR-10 dataset.
Even though the Triple GAN did not perform upto our expectations, this was a huge learning curve in the field of deep learning for me. This project
was funded by the IEEE NITK Student Chapter, of which I am a member. Code from the repository can be viewed here.
Custom Object Detection using YOLO with Darknet
I came up with a detailed tutorial on how one can use YOLO with Darknet for custom datasets. The darknet repository works out of the box
for some datasets like MS-COCO, but we can tune it to work for a dataset of our own. Due to a lack of clear documentation, I created a
step by step tutorial here.
FantasyOne - An F1 fantasy game
I'm a huge Formula 1 fan, and hence, my friend and I decided to create a web-based fantasy sports application for the sport. This
was a course project in college, and we learnt how to use NodeJS with express, managed a MySQL database and worked on front end design. I hope
to work on more such projects where my hobbies can intermix with my technical skills! Here is a link to the repository.
You can always view my GitHub or reach out to me to learn about some of the other projects I have worked on! In this section, I wanted to highlight projects that I find most important
and close to my heart.
