High-resolution tree counting and localization dataset in plain and hilly areas of eastern China

Trees are central organisms in maintaining global biodiversity and the health of the planet, and contribute extensively to biogeochemical cycles, and provide countless ecosystem services, including water quality control, wood stocks and carbon sequestration. Tree density is an important component of ecosystem structure, governing the rate of element processing and retention, as well as habitat suitability for many plant and animal species. The number of trees in a given area can also be a meaningful indicator to guide forest management practices and inform decision-making in public and governmental sectors. However, due to the complex distribution of trees, it has always been challenging to use remote sensing techniques to acquire it efficiently and effectively on a large spatial scale. In response to the growing need for individual trees scale research, we produce the Tree Counting Datasets based on GF-Ⅱ remote sensing images with a spatial resolution of 0.8m. This data set contains a total of 2400 samples in different geological scenarios in temperate and subtropical plains and hills, including wild woodland, urban and rural areas. Each sample pair consists of remote sensing images, tree annotation, and tree density maps generated by Gaussian convolution. The cross-validation experiment revealed the common counting networks could achieve the competitive performance (above 0.93) in terms of the determination coefficient (R²) between the ground truth and the estimated values and the average accuracy are greater than 84%. This dataset could be useful for tree density estimation, tree counting, and tree localization researches, thereby spurring biological analyses and facilitating model development for tasks that rely on individual tree prediction.