FruitsPark-Trunk10
收藏官方服务：
资源简介：
FruitsPark Dataset FruitsPark is a tree image dataset collected at a public fruit park in Japan. It contains 790 RGB images of 10 tree species captured from ground level using smartphones. The images were collected in different seasons and include variations in foliage, lighting, ground conditions, surrounding trees, and other background elements. The dataset is intended for research on tree species classification, background shortcut learning, and model robustness to seasonal and environmental changes. The dataset is divided into 580 training images and 210 validation images. Because no independent test set is provided, results obtained using the validation split should be clearly reported as validation performance.
提供机构：
Zenodo创建时间：
2026-07-16
