DeepSlide: Landslide Detection & Segmentation

DeepSlide is a high-performance, interactive web application built with Streamlit and PyTorch, designed to detect and segment landslide events from satellite imagery. It supports over 13 state-of-the-art deep learning segmentation backbones and architectures, enabling both single-model predictions and detailed multi-model comparison evaluations. The application allows users to easily upload custom TIFF or multispectral optical images and generate instant, high-resolution mask segmentations. By offering both a clean, streamlined single-model interface and a comprehensive side-by-side comparison layout, DeepSlide provides researchers and disaster response teams with a robust, scalable, and intuitive tool to evaluate prediction accuracy and mapping performance across various deep learning models simultaneously.