Projects Undertaken

Object detection using DOTA dataset and training using YOLOV8

View Document | View PPT

  • Detect objects such as airplanes in aerial images using the DOTA dataset and the YOLOv8s model.
  • A DOTA sub dataset selected converted from oriented bounding box (OBB) format to COCO format for compatibility with YOLOv8.
  • The YOLOv8 model trained on prepared dataset, preprocessed and added validation steps to improve detection performance.
  • The trained model detects objects in test images and the outputs visualized through an interactive Python-based interface.

Human-Elephant Conflict & Forest Fire Risk Assessment

View document

  • Applied Frequency Ratio & Weighted Overlay GIS modeling.
  • Studied vulnerability patterns in Muthanga Wildlife Range, Kerala.

Land Use Change Detection – Norcia, Italy

View document

  • Analyzed pre- and post-earthquake land patterns (2016 Italy earthquake).
  • Used remote sensing & GIS techniques for land-use classification.