---
dataset_info:
  features:
  - name: image
    dtype: image
  - name: label
    dtype: int64
  - name: antenna
    dtype: string
  - name: datetime
    dtype: string
  splits:
  - name: train
    num_bytes: 1833634770.762
    num_examples: 113409
  - name: validation
    num_bytes: 228715568.866
    num_examples: 14253
  - name: test
    num_bytes: 230174837.398
    num_examples: 14271
  download_size: 2352964096
  dataset_size: 2292525177.026
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: validation
    path: data/validation-*
  - split: test
    path: data/test-*
---

# e-Callisto Solar Flare Detection Dataset

![](https://www.fhnw.ch/en/++theme++web16theme/assets/media/img/university-applied-sciences-arts-northwestern-switzerland-fhnw-logo.svg)
[Institute of Data Science i4Ds, FHNW](https://i4ds.ch)  
Compiled by [Vincenzo Timmel | Kenfus](https://github.com/kenfus)

## Overview
This dataset comprises radio spectra from the [e-Callisto solar spectrometer network](https://www.e-callisto.org/index.html), annotated according to the [e-Callisto Label List by Christian Monstein](http://soleil.i4ds.ch/solarradio/data/BurstLists/2010-yyyy_Monstein/). It is designed for training machine learning models to automatically detect and classify solar flares, using data collected via the [ecallisto_ng Package](https://github.com/i4Ds/ecallisto_ng).

## Non Radio-Sunburst Images
For every Radio-Sunburst image, five Non-Sunburst images are included (Label 0).

## Data Collection
The dataset encompasses observations from several antennas, each documenting specific periods of data collection. Below is the updated list of stations with their respective data collection ranges:

| Antenna               | Min Date   | Max Date   |
|-----------------------|------------|------------|
| ALASKA-COHOE_63       | 2022-04-09 | 2024-02-22 |
| ALASKA-HAARP_62       | 2021-11-18 | 2024-02-23 |
| ALGERIA-CRAAG_59      | 2021-04-23 | 2024-02-22 |
| ALMATY_58             | 2021-02-28 | 2024-02-23 |
| AUSTRIA-UNIGRAZ_01    | 2021-01-20 | 2024-02-23 |
| Australia-ASSA_02     | 2021-02-13 | 2021-12-09 |
| Australia-ASSA_62     | 2021-12-10 | 2024-02-22 |
| BIR_01                | 2021-04-17 | 2024-02-14 |
| EGYPT-Alexandria_02   | 2021-08-20 | 2024-02-21 |
| GERMANY-DLR_63        | 2022-11-11 | 2024-02-22 |
| GLASGOW_01            | 2022-01-07 | 2024-02-22 |
| HUMAIN_59             | 2021-01-20 | 2024-02-23 |
| INDIA-GAURI_01        | 2022-04-20 | 2024-02-21 |
| INDIA-OOTY_02         | 2021-12-09 | 2024-02-23 |
| KASI_59               | 2021-04-22 | 2024-02-23 |
| MEXART_59             | 2021-02-18 | 2024-02-21 |
| MEXICO-FCFM-UANL_01   | 2023-09-02 | 2024-02-21 |
| MEXICO-LANCE-B_62     | 2022-03-30 | 2022-08-02 |
| MONGOLIA-UB_01        | 2021-03-01 | 2024-02-16 |
| MRO_59                | 2021-03-01 | 2024-02-16 |
| MRO_61                | 2021-02-28 | 2024-02-16 |
| NORWAY-EGERSUND_01    | 2022-10-16 | 2024-02-23 |
| SSRT_59               | 2022-10-30 | 2024-02-23 |
| SWISS-Landschlacht_62 | 2021-10-05 | 2024-02-23 |
| TRIEST_57             | 2021-01-20 | 2024-02-23 |
| USA-ARIZONA-ERAU_01   | 2022-05-15 | 2024-02-22 |


## Data Augmentation
Data augmentation is applied by subtracting random minutes before the start of a detected radio sunburst, thereby generating 15-minute images that include the onset of a radio-sunburst.

## Caution
Preprocessing, including label cleanup based on specific assumptions, has been applied to the dataset. Users should note that the labels might not be entirely accurate, reflecting potential inaccuracies

## Distribution
**Train**

![image/png](https://cdn-uploads.huggingface.co/production/uploads/6564a6f4ff73855b0327455a/cis_YvRe1OQhdtWIesl1M.png)

**Validation**

![image/png](https://cdn-uploads.huggingface.co/production/uploads/6564a6f4ff73855b0327455a/04VONOgKHOJmrqTzeV2y3.png)

**Test**

![image/png](https://cdn-uploads.huggingface.co/production/uploads/6564a6f4ff73855b0327455a/ct8CG83ryHsvhgtRrMXmJ.png)