BarkVisionAI: Novel dataset for rapid tree species identification

Area of Science:

• Botany
• Computer Science
• Ecology

Background:

• Accurate tree species identification is vital for forest management and conservation.
• Ground-level bark images offer rich data but existing datasets lack diversity.
• Limitations in current tree bark datasets hinder real-world applications.

Purpose of the Study:

• To introduce BarkVisionAI, a comprehensive dataset of tree bark images.
• To address the limitations of existing datasets in terms of species diversity and temporal attributes.
• To facilitate research in automated tree species identification and ecological monitoring.

Main Methods:

• Collected 156,001 tree bark images across 13 species from diverse Indian forest types.
• Labeled each image with location, species, device attributes, and timestamp.
• Conducted benchmarking experiments using standard image classification models.

Main Results:

• BarkVisionAI comprises 156,001 labeled images for 13 tree species.
• The dataset includes detailed metadata, enabling robust analysis.
• Benchmarking demonstrated the dataset's effectiveness for machine learning models.

Conclusions:

• BarkVisionAI is a valuable resource for automated tree species identification.
• The dataset supports ecological research and environmental monitoring.
• It facilitates the development of reliable, real-world applications for forestry.