Oil Palm Ganoderma Disease Dataset

Opening new doors of possibilities in agriculture alternative

In an effort to modernize the agriculture sector in Malaysia, we have decided to make-public a hyperspectral image dataset of oil palm plants. This was made possible in collaboration with FGV R&D Sdn Bhd that allowed the collection of hyperspectral image data at their specific test sites. This activity was also partially supported by the Center for Remote Sensing and Surveillance Technologies (CRSST), Multimedia University. The hyperspectral images in the dataset have a size of 1024×1024 and consist of 79 bands with a wavelength range of 510 nm - 900 nm.

The dataset contains 10 hyperspectral images with a total of 75 oil palm plants highlighted with a bounding box. Each bounding box is labelled as the stage of Ganoderma disease which the plant is infected with. There is a total of four Ganoderma stages in the dataset: H (Healthy), A (Early Stage), B (Middle Stage), and C (Final Stage).


The dataset is available for download.

Sneak Peek

# Class Count
0 Healthy 59
1 Stage A Ganoderma 7
2 Stage B Ganoderma 4
3 Stage C Ganoderma 5

The figures below are some samples of the hyperspectral images in the dataset displayed in their RGB bands (0, 9, 18) with their respective bounding boxes.