Bark Texture Recognition of Indian Trees: Augmented Dataset for Robust Classification

The bark of a tree provides stable and distinctive visual patterns that play a crucial role in automated tree species identification. Our previously released dataset, Bark Texture Recognition of Indian Trees: A Bangalore-Centric Dataset, included 558 RGB images representing 22 commonly found tree species in Karnataka, India. In this extended version, we present an augmented and expanded dataset designed to improve robustness, generalization, and model performance for texture-based classification tasks. The dataset incorporates multiple augmentation strategies, including rotation, brightness variation, contrast adjustments, scaling, and noise addition, effectively increasing intra-class variability while preserving the natural characteristics of bark textures. This enhanced version supports deep learning, few-shot learning, and transfer learning experiments by providing richer variations that simulate real-world environmental conditions. The extended dataset aims to address challenges such as limited sample size, illumination differences, viewpoint variations, and bark texture diversity, thereby offering a more comprehensive resource for researchers working in forestry, biodiversity conservation, and computer vision-based species identification.