Celyrix 1.0 Vision
Satellite Imagery Classification
1 min readupdated
Celyrix 1.0 Vision is a lightweight computer vision model by Mirutex Technologies designed to analyze RGB satellite imagery and classify it into five land cover categories. Built on TensorFlow and Keras, the model focuses on efficient inference on standard hardware and reproducible research. It classifies 64×64 pixel images into forests, rivers, residential areas, agricultural land, and industrial sites. It is well-suited for researchers, geospatial developers, and educational experiments using EuroSAT data.
Key features
- Satellite vision: classification of RGB imagery into five land cover categories.
- Lightweight architecture: efficient inference on standard computing hardware.
- Reproducible research: documented training and deployment pipelines built on TensorFlow and Keras.
- EuroSAT dataset: trained on selected classes from the popular Sentinel-2 satellite dataset.
What you can do
- Imagery analysis: integrate the model into satellite data processing pipelines for automated land categorization.
- Geospatial research: use it to classify forests, rivers, residential, and industrial areas in academic projects.
Example
Loading the model repository using huggingface_hub:
from huggingface_hub import snapshot_download
model_dir = snapshot_download(
repo_id="Mirutex/Celyrix-1.0-Vision"
)
print(model_dir)Read more
DetailsArchitecture: Convolutional Neural Network
Framework: TensorFlow / Keras
Input data: 64×64 pixel RGB imagery
Categories: Forest, River, Residential, Agriculture, Industrial
License: PolyForm Noncommercial 1.0.0










Comments