Indian Flora Project

Performed transfer learning on various CNN models using our new Indic-Leaf dataset of 112 Indian plants for identification and disease detection with results showing Top-1 and Top-5 precision of 90.08 and 96.90

Deployed auto-segmenter, auto-recommender systems and a collaborative web-portal using Angular and NodeJS and an Android app where users can add new observations and query for species identification.

Advisor: C. V. Jawahar