Documents

Building a Satellite Image Classifier with Streamlit and 3D Viz

4 min readMay 22, 2025Feb 22, 2026

App Project

This is a satellite image classification visualization app built with Streamlit. Development ran from roughly May 22–26, 2025.

App Structure

flowchart TD
    A[Streamlit App] --> B[이미지 업로드]
    B --> C[ResNet50 분류]
    C --> D{결과 시각화}
    D --> E[3D 지형 맵]
    D --> F[신뢰도 차트]
    D --> G[카테고리 피봇]

Tech Stack

  • Streamlit: Web UI framework
  • PyTorch: ResNet50 model
  • Plotly: 3D visualization
  • Pillow: Image preprocessing

Fast Development

Shipped in 5 days with 13 commits. It was a prototype that leaned heavily on Streamlit's strengths.

Streamlit: Pros and Cons

Pros:

  • Minimal UI code
  • Rapid prototyping
  • Auto-reload

Cons:

  • Complex state management
  • Limited customization
  • Not suited for large-scale apps

It's a simple toy project, but it was enough to get reasonably comfortable with Streamlit.

Tags
Streamlitsatellite imageryclassification3DPyTorch