At the Centre of Excellence for Data Science (CEDS), IEM Kolkata, I contribute to faculty-mentored research on computer vision and deep learning for autonomous vehicle perception. I use PyTorch to benchmark semantic segmentation architectures, comparing classical Fully Convolutional Networks with transformer models.
For my Road Segmentation for Autonomous Driving project, I fine-tuned an ADE20K-pretrained SegFormer-B2, achieving 92.95% validation IoU and a 96.34% Dice Score. I’ve also developed projects in citation verification, dietary label analysis, and computer vision-based posture detection.

