LiFE-BA-Net: A Lightweight Network for Depth Feature Enhancement and Boundary-Aware in Fire-Front Segmentation

The UAV_dataset in this demo follows the official VOC dataset format. Within the ImageSets directory, text files specifying the splits for the semantic segmentation dataset are provided. Specifically, the Segmentation subdirectory contains train.txt, val.txt, and test.txt, which correspond to the training, validation, and test sets, respectively, facilitating convenient data preprocessing and dataset partitioning. (1) The JPEGImages directory contains the images used for training, validation, and testing. Each image filename corresponds to the entries listed in train.txt, val.txt, and test.txt within the Segmentation subdirectory of ImageSets, thereby defining the training, validation, and test splits. (2) The SegmentationClass directory contains the labels for training, validation, and testing. Each label filename corresponds to the entries in train.txt, val.txt, and test.txt in the Segmentation subdirectory of ImageSets, ensuring consistency in the assignment of images and labels across the dataset splits.