Botanicum: a Telegram Bot for Tree Classification

Autor: Daria Korotaeva, Maksim Khlopotov, Anastasiia Makarenko, Ekaterina Chikshova, Natalia Startseva, Anastasiia Chemysheva
Jazyk: angličtina
Rok vydání: 2018
Předmět:
Zdroj: Proceedings of the XXth Conference of Open Innovations Association FRUCT, Vol 426, Iss 22, Pp 88-93 (2018)
Druh dokumentu: article
ISSN: 2305-7254
2343-0737
DOI: 10.23919/FRUCT.2018.8468278
Popis: The article presents Botanicum - a telegram bot for tree classification based on the image of a leaf. In this work, we used our previously developed classification model which underlies the bot. Botanicum can identify 20 different tree species typical for Russian flora. The process of user interaction includes a simple chat and a short instruction on how to take a photo. After image verification, the bot either outputs error description or the classification results. We managed to increase the classification accuracy to 97,8% by outputting top results instead of one single option.
Databáze: Directory of Open Access Journals