BambooBD: Classification of Bangladesh’s Bamboo varieties with an extensive Image Dataset
Description
This dataset was developed to support automated, image-based classification of bamboo varieties as a scalable alternative to traditional morphological identification, which is time-consuming and requires specialized botanical expertise. The underlying hypothesis is that distinguishing visual features of bamboo culms and grid structures such as color, texture, and surface patterns, are sufficiently distinct across species to enable reliable classification using deep learning, even under varying environmental conditions (lighting, season, weather). The dataset contains high-resolution images of nine bamboo varieties cultivated in Bangladesh, collected at the Bangladesh Forest Research Institute (BFRI), Chittagong, between October 2024 and January 2025: Bambusa cacharensis (Betua), Dendrocalamus latiflorus (Chaina), Bambusa ventricosa (Ghati), Bambusa sp. (Hedge), Bambusa bambos (Kanta), Bambusa tulda (Mritinga), Bambusa comillensis (Sine Kais), Bambusa vulgaris var. striata (Sharna), and Dendrocalamus giganteus (Bhudum). Images were captured using multiple smartphone devices (iPhone 15 Pro Max, iPhone 11, Redmi Note 9 Pro, and Realme 8) under natural, varied lighting and seasonal conditions to reflect real-world variability rather than controlled studio settings. All images were standardized to 512×512 pixels at 96 dpi for computational consistency. The dataset comprises 1,960 original images, expanded to 9,800 total images through data augmentation (random rotation, horizontal flipping, brightness adjustment, and random cropping). The dataset is organized into a main folder, "BambooBD," with two subfolders : 1. Bamboo Grids Original Images and 2. Bamboo Grids Augmented Images, each divided into nine species-specific folders labeled by native name. Researchers can use this dataset to train and benchmark image classification models, study morphological variation across bamboo species, support cross-breeding and genetic diversity research, and inform agroforestry and biodiversity conservation efforts specific to Bangladesh.
Files
Institutions
- International Islamic University ChittagongChittagong, Chittagong