TreeMortNet Training and Validation Data

This repository contains four NumPy arrays, each used to train and validate TreeMortNet, a residual convolutional neural network that predicts individual tree-level mortality using National Agriculture Inventory Program (NAIP) imagery and lidar-delineated tree crown polygons:X_train.npy: Four-band (RGB + NIR) image chips, masked to the extent of tree crown polygons and rescaled to 64x64 pixels, with pixel values ranging from 0-255. This array contains image data from 86715 trees, distributed throughout the Western US, used in model training.y_train.npy: Label data for each of the 86715 trees in the training dataset, where 0 = live and 1 = dead.X_val.npy: The equivalent of X_train.npy, but for validation data, with 22196 trees.y_val.npy: The equivalent of y_train.npy, but for validation data, with 22196 trees.