Treevill: National Botanical Garden Unique & Rare Tree Argument Dataset

The Treevill: National Botanical Garden Unique & Rare Tree Argument Dataset was meticulously curated to provide high-quality, diverse images of tree species found in the National Botanical Garden of Bangladesh (N.B.G.). The dataset is designed to support a wide range of research and machine learning applications, particularly in biodiversity studies, image classification, and object recognition.

Source of Data:

The images in this dataset were collected directly from the National Botanical Garden of Bangladesh, which houses a wide variety of unique and rare tree species.
Data collection was performed under natural conditions to ensure the images captured a diverse representation of each species from multiple angles, lighting conditions, and environmental settings.
Species Documentation:

Data Augmentation Process:
Grayscale: Images were converted to black-and-white to focus more on the structural features of the trees, reducing the influence of color.

Posterized: The number of colors in the image was reduced to highlight shapes and contrasts, simulating the effect of various imaging techniques.

Rotated: Images were randomly rotated to simulate different viewpoints and perspectives, enabling the model to better recognize the tree species from any angle.

Flipped Horizontal/Vertical: Horizontal and vertical mirroring was applied to simulate alternative tree orientations, increasing the variety in the dataset.

Bright/Contrast Adjustments: The brightness and contrast of the images were modified to simulate varying environmental lighting conditions, such as sunny, cloudy, or shaded areas.

Blurred: Some images were intentionally blurred to simulate focus variations and to mimic different depth-of-field effects in real-world photography.

Edge Enhancement: Sharpness was applied to highlight edges and features of the trees, making fine details stand out.

Inverted: Colors in some images were inverted to test the model's robustness against unusual or extreme lighting conditions.

Thumbnail: Lower-resolution versions of images were created to emphasize larger patterns and visual structures, useful for testing model performance at lower resolutions.

Cropped: Images were cropped to simulate partial views or closer inspections of specific parts of the tree, focusing on details such as leaves, bark, or fruits.

Sharpened: The sharpness of some images was enhanced to emphasize finer details of the tree species and their distinguishing features.

Sepia: A sepia tone was applied to simulate aged or vintage photographs, which may be encountered in some botanical archives or historical documents.