Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -1,71 +1,78 @@
|
|
| 1 |
---
|
| 2 |
license: gpl-2.0
|
| 3 |
-
library_name: pytorch
|
| 4 |
tags:
|
| 5 |
-
-
|
| 6 |
-
-
|
| 7 |
-
-
|
| 8 |
-
-
|
| 9 |
-
-
|
| 10 |
-
-
|
| 11 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
---
|
| 13 |
|
| 14 |
-
#
|
| 15 |
-
|
| 16 |
-
**FRTSearch** is a deep learning model for detecting **Pulsars**, **RRATs**, and **Fast Radio Bursts (FRBs)** in radio astronomical observation data.
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
## Model Details
|
| 20 |
|
| 21 |
-
-
|
| 22 |
-
- **Framework**: PyTorch
|
| 23 |
-
- **Task**: Instance Segmentation / Object Detection
|
| 24 |
-
- **License**: GPL-2.0
|
| 25 |
|
|
|
|
| 26 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 27 |
|
| 28 |
-
##
|
| 29 |
|
| 30 |
```python
|
| 31 |
-
import
|
| 32 |
-
from mmdet.apis import init_detector
|
| 33 |
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
```
|
| 39 |
|
| 40 |
-
|
| 41 |
|
|
|
|
| 42 |
|
|
|
|
| 43 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 44 |
|
| 45 |
-
|
| 46 |
-
| File | Description |
|
| 47 |
-
|------|-------------|
|
| 48 |
-
| `models/hrnet_epoch_36.pth` | Model weights (Mask R-CNN with HRNet backbone) |
|
| 49 |
-
| `configs/detector_FAST.py` | Configuration for FAST telescope data |
|
| 50 |
-
| `configs/detector_SKA.py` | Configuration for SKA telescope data |
|
| 51 |
-
|
| 52 |
-
## Citation
|
| 53 |
|
| 54 |
```bibtex
|
| 55 |
@article{zhang2026frtsearch,
|
| 56 |
-
title={FRTSearch: Unified Detection and Parameter Inference of Fast Radio Transients using Instance Segmentation},
|
| 57 |
-
author={Zhang, Bin and Wang, Yabiao and Xie, Xiaoyao et al.},
|
| 58 |
-
year={2026}
|
| 59 |
}
|
| 60 |
```
|
| 61 |
|
| 62 |
-
##
|
|
|
|
|
|
|
| 63 |
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
- **Dataset**: [CRAFTS-FRT](https://doi.org/10.57760/sciencedb.Fastro.00038)
|
| 67 |
|
| 68 |
-
##
|
| 69 |
|
|
|
|
|
|
|
|
|
|
| 70 |
|
| 71 |
-
|
|
|
|
| 1 |
---
|
| 2 |
license: gpl-2.0
|
|
|
|
| 3 |
tags:
|
| 4 |
+
- Radio Astronomy
|
| 5 |
+
- Mask-RCNN
|
| 6 |
+
- Pulsar
|
| 7 |
+
- RRAT
|
| 8 |
+
- FRB
|
| 9 |
+
- Signal Processing
|
| 10 |
+
library_name: pytorch
|
| 11 |
+
pipeline_tag: image-segmentation
|
| 12 |
+
datasets:
|
| 13 |
+
- CRAFTS-FRT
|
| 14 |
+
metrics:
|
| 15 |
+
- recall
|
| 16 |
+
- FPR
|
| 17 |
---
|
| 18 |
|
| 19 |
+
# FRTSEARCH: FAST RADIO TRANSIENT SEARCH
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
|
| 21 |
+
FRTSearch is an end-to-end deep learning framework for detecting and characterizing Fast Radio Transients (FRTs), including: Pulsars, Rotating Radio Transients (RRATs) and Fast Radio Bursts (FRBs).
|
|
|
|
|
|
|
|
|
|
| 22 |
|
| 23 |
+
## MODEL INFO
|
| 24 |
|
| 25 |
+
| Item | Value |
|
| 26 |
+
| :--- | :--- |
|
| 27 |
+
| Backbone | HRNet-W32 |
|
| 28 |
+
| Input | 256 × 8192 (freq × time) |
|
| 29 |
+
| Size | 381.75 MB |
|
| 30 |
+
| Formats | .fits (PSRFITS), .fil (Filterbank) |
|
| 31 |
+
| Bit Depth | 1/2/4/8/32-bit |
|
| 32 |
|
| 33 |
+
## QUICK START
|
| 34 |
|
| 35 |
```python
|
| 36 |
+
from huggingface_hub import hf_hub_download
|
|
|
|
| 37 |
|
| 38 |
+
model_path = hf_hub_download(
|
| 39 |
+
repo_id="waterfall109/FRTSearch",
|
| 40 |
+
filename="models/hrnet_epoch_36.pth"
|
| 41 |
+
)
|
| 42 |
```
|
| 43 |
|
| 44 |
+
Or download directly from Files and versions.
|
| 45 |
|
| 46 |
+
## TEST SAMPLES
|
| 47 |
|
| 48 |
+
This repository includes 5 test samples from 2 different telescopes to demonstrate cross-facility performance:
|
| 49 |
|
| 50 |
+
| Telescope | FRB | DM (pc cm⁻³) |
|
| 51 |
+
| :--- | :--- | :--- |
|
| 52 |
+
| FAST | 20121102, 20180301, 20201124 | 565, 420, 525 |
|
| 53 |
+
| ASKAP | 20180119, 20180212 | 400, 168 |
|
| 54 |
|
| 55 |
+
## CITATION
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 56 |
|
| 57 |
```bibtex
|
| 58 |
@article{zhang2026frtsearch,
|
| 59 |
+
title={FRTSearch: Unified Detection and Parameter Inference of Fast Radio Transients using Instance Segmentation},
|
| 60 |
+
author={Zhang, Bin and Wang, Yabiao and Xie, Xiaoyao et al.},
|
| 61 |
+
year={2026}
|
| 62 |
}
|
| 63 |
```
|
| 64 |
|
| 65 |
+
## TEST SAMPLE REFERENCES
|
| 66 |
+
|
| 67 |
+
When using the test samples, please also cite the original observations:
|
| 68 |
|
| 69 |
+
* FAST samples: Guo et al. (2025)
|
| 70 |
+
* SKA samples: Shannon et al. (2018)
|
|
|
|
| 71 |
|
| 72 |
+
## LICENSE & ACKNOWLEDGMENTS
|
| 73 |
|
| 74 |
+
GPL-2.0 | Based on MMDetection & PRESTO
|
| 75 |
+
|
| 76 |
+
---
|
| 77 |
|
| 78 |
+
🌌 Questions? GitHub Issues
|