Reseach Assistant (SPARK Fellowship)
I was awarded the coveted SPARK fellowship to pursue a research
internship at IIT Roorkee under Dr. Sparsh Mittal (Assistant Prof, Dept. Electrical Engineering, IIT Roorkee). My work involved going through research papers to design novel architectures to achieve higher mean average precision (mAP) in object detection. I used a single shot detector (SSD) for object detection. I designed different models to increase the mAP with a minor increase in inference time. My main research area was centered around using various SSD architecture plugins to make it more scalable and easy to improve.
My idea of using feature-rich plugins with the model helped to increase the mAP(accuracy) by 3%. As a result of this, I published my work at the 9th IAPR TC3 Workshop on Artificial Neural Networks in Pattern Recognition, ANNPR 2020. I then worked on another research paper
involving deep learning for object detection on aerial images from drones. The research paper was submitted to a reputed journal and is currently under review.
Internship duration: 3 months
Skills: Deep Learning, Object Detection
Deep Learning Framework: PyTorch