Dataset for Forest Object Segmentation

This dataset is designed for experiments in segmenting forest objects and consists of horizontal and vertical images, including some winter images. The images were provided by Komatsu and have been manually selected and annotated by humans. Training Dataset 1: 150 horizontal and 150 vertical images Horizontal Images: Komatsu Halvattnet drone video in forest, Forest photos Tribladsberget VindelnVertical Images: Komatsu Photos from ground, Tribladsberget Vindeln.Validation Dataset 1: 30 horizontal images Validation Dataset 2: 30 vertical images Images were authored by KOMATSU. If you wish to use them please contact [e-mail] 

Masks were authored by DAC. digital. If you wish to use them please contact [e-mail]

本数据集专为森林目标分割实验设计，涵盖横向与纵向图像，其中包含部分冬季影像。所有图像均由小松（Komatsu）提供，并经人工筛选与标注。

训练数据集1：包含150张横向图像及150张纵向图像。其中横向图像取自小松公司于Tribladsberget Vindeln林区的Halvattnet无人机航拍视频，以及该林区的森林实拍照片；纵向图像则为小松公司在Tribladsberget Vindeln林区获取的地面实拍照片。

验证数据集1：包含30张横向图像。
验证数据集2：包含30张纵向图像。

本数据集的图像版权归小松（KOMATSU）所有，若需使用该图像资源，请联系邮箱[e-mail]。

本数据集的图像掩码标注由DAC. digital制作，若需使用该标注资源，请联系邮箱[e-mail]。