BongFloralpedia: A comprehensive collection of diverse types of flowers growing in fields of Ranaghat, Bengal

This database is a collection of a total of 1920 images of flowers
belonging to nine different types of flower species namely, Tuberose (247 images), Marigold
(206 images), Ixora coccinea (202 images), Hibiscus (208 images), Crape jasmine (202
images), Dhatura (200 images), Bougainvillea (207 images), 4o’ clock (208 images) and Rose
(240 images) captured from real-fields located in Ranaghat, West Bengal, India.
The images included in this database comprises of complex backgrounds containing weeds,
soil, parts of other plants, etc. This database is created with the aim of providing a
comprehensive collection of labelled floral images to train deep neural networks to perform
automated classification of nine diverse types of floral species. The inclusion of flower images
possessing complex backgrounds in this database enables it to train deep neural networks for
performing highly accurate real-life identification of floral species.
The images incorporated in this database are captured using mobile phones, Samsung Galaxy
S20 FE 5G (Camera specifications: 12MP (Dual Pixel) OIS F 1.8 Wide Rear Camera+ 8MP
OIS Tele Camera + 12 MP Ultra Wide 30X Space Zoom) and Redmi Note 9 pro max (Camera
specifications: AI Quad camera array 64MP+8MP+5MP+2MP).
The images of flowers included in this database are captured from real-fields located in
Ranaghat, West Bengal, India which is one of the most well-known flower cultivation areas of
the state, so the name of this database is termed as BongFloralpedia considering the place of
origin of flowers whose images are included in this database.

本数据库共计收录1920张花卉图像，涵盖9个不同的花卉物种，分别为晚香玉（Tuberose，247张）、万寿菊（Marigold，206张）、龙船花（Ixora coccinea，202张）、木槿（Hibiscus，208张）、狗牙花（Crape jasmine，202张）、曼陀罗（Dhatura，200张）、叶子花（Bougainvillea，207张）、紫茉莉（4o’ clock，208张）以及玫瑰（Rose，240张），所有图像均采集自印度西孟加拉邦拉纳哈特的田间实拍场景。
本数据库所收录的图像均带有复杂背景，包含杂草、土壤、其他植物的部分区域等。本数据库的构建目标是提供一套完整的带标注花卉图像集，用于训练深度神经网络以实现9类不同花卉物种的自动化分类。由于收录了带有复杂背景的花卉图像，该数据库可用于训练深度神经网络，实现高精度的现实场景花卉物种识别。
本数据库中的图像均通过两款移动设备采集：三星Galaxy S20 FE 5G（摄像头参数：1200万像素（双像素）光学防抖F1.8广角后置摄像头+800万像素光学防抖长焦摄像头+1200万像素超广角镜头，支持30倍空间变焦），以及Redmi Note 9 pro max（摄像头参数：AI四摄阵列，6400万像素+800万像素+500万像素+200万像素）。
本数据库收录的花卉图像均采集自印度西孟加拉邦拉纳哈特的田间实拍场景，该区域为该州最知名的花卉种植基地之一。鉴于收录花卉图像的原产地属性，本数据库被命名为BongFloralpedia。