README.md

metadata

title: GeoSpace Drone Intelligence
emoji: 🛰️
colorFrom: blue
colorTo: green
sdk: streamlit
sdk_version: 1.32.0
app_file: app.py
pinned: false

🛰️ GeoSpace — Two-Stage Drone Intelligence Platform

A professional geospatial AI pipeline for analyzing drone and aerial imagery.

Two-Stage Pipeline

Stage 1 — Land Cover Segmentation (SegFormer-B2) Pixel-level segmentation detecting: Buildings, Roads, Vegetation, Water, Bare Land, Forest, Farmland, Sand/Desert

Stage 2 — Landform & Geological Classification (SigLIP Satellite Classifier) Each detected region is independently classified into landform types with geological context inference.

Three Input Modes

• 📷 Image — Upload any drone/aerial image
• 🎥 Video — Upload drone footage, get annotated output video
• 📡 Live Feed — Webcam-based live drone feed simulation

Models Used

• nvidia/segformer-b2-finetuned-ade-512-512
• prithivMLmods/SAT-Landforms-Classifier

Tech Stack

Transformers · PyTorch · OpenCV · Streamlit