https://github.com/i4ds/flaresense-v2
https://github.com/i4ds/flaresense-v2
Last synced: 4 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/i4ds/flaresense-v2
- Owner: i4Ds
- Created: 2024-02-29T16:25:15.000Z (over 2 years ago)
- Default Branch: main
- Last Pushed: 2025-12-18T16:53:02.000Z (6 months ago)
- Last Synced: 2025-12-21T20:37:37.040Z (6 months ago)
- Language: HTML
- Size: 21 MB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# FlareSense v2
FlareSense is an experimental project that trains a convolutional neural network
on e-Callisto radio spectrograms to detect solar radio bursts. The repository
contains training code, prediction utilities, a web app for user interaction,
and several notebooks used during development.
## Installation
The code requires **Python 3.11**. Use the conda environment for dependencies:
```bash
conda activate flaresense-v2
pip install -r requirements.txt
```
## Training
Training is driven by YAML configuration files. A typical run looks like:
```bash
python main.py --config configs/best_v2.yml
```
The `main.sh` script shows how to submit a job on a SLURM cluster.
## Prediction
Run inference on a dataset with:
```bash
python pred_dataset.py
```
For live prediction through a Gradio interface execute:
```bash
python pred_live.py
```
## Web App
The web app in `app.py` provides a user interface for uploading data, running predictions, and viewing results. It is built with Flask and integrates with the prediction models.
To hide the complete path, we make a symlink between where the images are and where they are shown.
## Evaluation
To reproduce our results, run the following command:
```bash
python main.py --config configs/best_v2.yml
```
## Notebooks
All notebooks can be found in the `_notebooks` directory. They provide
exploratory data analysis, model investigations, and visualizations.
## Deployment
FlareSense is deployed as Linux systemd services for production use:
- **flaresense_app.service**: Manages the web app (`app.py`) for user-facing interactions.
- **flaresense.service**: Handles continuous prediction tasks.
## Service Management
If FlareSense is deployed as systemd services, you can inspect the logs with:
```bash
sudo journalctl -u flaresense_app.service # For web app logs
sudo journalctl -u flaresense.service # For prediction service logs
```
After modifying the service files, redeploy with:
```bash
sudo systemctl restart flaresense_app.service
sudo systemctl restart flaresense.service
sudo systemctl daemon-reload
```
The .serice-file can be found here: /etc/systemd/system/flaresense_app.service
---
This repository is provided for reference without any warranty.