Voxel51/FinnWoodlands

FinnWoodlands是一个专为林业和机器人应用设计的森林数据集。它包含RGB立体图像、点云和稀疏深度图，以及语义、实例和全景分割的地面真实手动注释。数据集包含4,226个手动注释的对象，其中2,562个对象（60.6%）对应于树干，分为三个实例类别：云杉树、桦树和松树。其他注释包括作为实例对象的障碍物和语义stuff类：湖泊、地面和轨道。数据集由Tampere大学的研究人员创建，用于森林环境中的语义分割、实例分割等应用。

FinnWoodlands is a forest dataset designed for forestry and robotics applications. It consists of RGB stereo images, point clouds, and sparse depth maps, along with ground truth manual annotations for semantic, instance, and panoptic segmentation. The dataset comprises 4,226 manually annotated objects, with 2,562 objects (60.6%) corresponding to tree trunks classified into three instance categories: Spruce Tree, Birch Tree, and Pine Tree. Additional annotations include Obstacles as instance objects and semantic stuff classes: Lake, Ground, and Track. The dataset was created by researchers at Tampere University for applications such as semantic segmentation and instance segmentation in forest environments.