Indoor Plant Disease Image Dataset for Classification
Description
This dataset contains indoor plant leaf images developed for research on automated plant disease classification and data augmentation. The dataset was created as part of the study “Deep Learning-Driven Leaf Disease Classification in Indoor Plants with YOLOv11: Stable Diffusion Augmentation Approach,” presented at the 2025 28th International Conference on Computer and Information Technology (ICCIT). The original dataset consists of 787 high-resolution images covering nine classes across three commonly cultivated indoor plant species: Money Plant, Snake Plant, and Spider Plant. The classes include Money Plant Bacterial Wilt Disease, Money Plant Healthy, Money Plant Manganese Toxicity, Snake Plant Anthracnose, Snake Plant Healthy, Snake Plant Leaf Withering, Spider Plant Fungal Leaf Spot, Spider Plant Healthy, and Spider Plant Leaf Tip Necrosis. Images were manually collected from several plant nurseries in Sylhet City, Bangladesh, under natural lighting using three different smartphones. The dataset is provided in three separate collections: 1. 01_raw_dataset_787_images.zip - The original dataset containing 787 manually collected images. 2. 02_traditional_augmented_dataset.zip - Images generated using traditional geometric and photometric augmentation with the Albumentations library, including horizontal flipping, random 90-degree rotation, brightness and contrast adjustment, color jitter, and Gaussian blur. This process expanded the dataset from 787 to 4,722 images. 3. 03_diffusion_augmented_dataset.zip - Synthetic image variants generated using Stable Diffusion v1.5 in image-to-image mode. Class-specific text prompts were used with the original images as inputs. The generation process used a denoising strength of 0.1, guidance scale of 7.5, and 30 inference steps. Generated images were manually reviewed for diagnostic relevance and visual realism. During preprocessing, images were quality-checked, converted to JPEG, resized to 224 × 224 pixels, and manually cropped to isolate the leaf region and reduce irrelevant background information. This dataset is intended for research in indoor plant disease classification, computer vision, deep learning, image augmentation, synthetic data generation, and plant health monitoring. Associated publication: Nishad Mahmud Opu, Md Saiful Islam, Sayed Hanzala Abdullah, and Ruma Akter, “Deep Learning-Driven Leaf Disease Classification in Indoor Plants with YOLOv11: Stable Diffusion Augmentation Approach,” 2025 28th International Conference on Computer and Information Technology (ICCIT), pp. 443-448, 2025. DOI: 10.1109/ICCIT68739.2025.11491324. Researchers using this dataset are encouraged to cite the associated publication.
Files
Institutions
- Sylhet Engineering CollegeSylhet Division, Sylhet