Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,84 @@
|
|
| 1 |
-
---
|
| 2 |
-
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- en
|
| 4 |
+
license: cc-by-4.0
|
| 5 |
+
pretty_name: image-impeccable
|
| 6 |
+
---
|
| 7 |
+
|
| 8 |
+
# Dataset Card for Image Impeccable
|
| 9 |
+
|
| 10 |
+
## Dataset Description
|
| 11 |
+
|
| 12 |
+
This data was produced by ThinkOnward for the [Image Impeccable Challenge](https://thinkonward.com/app/c/challenges/image-impeccable), using a synthetic seismic dataset generator called [Synthoseis](https://github.com/sede-open/synthoseis).
|
| 13 |
+
|
| 14 |
+

|
| 15 |
+
|
| 16 |
+
- **Created by:** Mike McIntire and Jesse Pisel
|
| 17 |
+
- **License:** CC 4.0
|
| 18 |
+
|
| 19 |
+
## Uses
|
| 20 |
+
|
| 21 |
+
### How to generate a dataset
|
| 22 |
+
|
| 23 |
+
This dataset is provided as paired noisy and clean seismic volumes. Follow the following step to load the data to `numpy` volumes
|
| 24 |
+
|
| 25 |
+
```python
|
| 26 |
+
import pandas as pd
|
| 27 |
+
import numpy as np
|
| 28 |
+
def parquet2array(parquet_file, original_shape=(1259,300,300)):
|
| 29 |
+
df = pd.read_parquet(parquet_file)
|
| 30 |
+
data_only = df.drop(columns=['Row', 'Col'])
|
| 31 |
+
# Convert the DataFrame back to a 2D numpy array
|
| 32 |
+
reshaped_array = data_only.values
|
| 33 |
+
# Reshape the 2D array back into a 3D array
|
| 34 |
+
array = reshaped_array.reshape(original_shape)
|
| 35 |
+
return array
|
| 36 |
+
```
|
| 37 |
+
|
| 38 |
+
## Dataset Structure
|
| 39 |
+
|
| 40 |
+
- train (250 paired noisy and clean volumes)
|
| 41 |
+
- 42570686
|
| 42 |
+
- seismicCubes_RFC_fullstack_2024.42570686.parquet
|
| 43 |
+
- seismic_w_noise_vol_42570686.parquet
|
| 44 |
+
- 42576639
|
| 45 |
+
- seismicCubes_RFC_fullstack_2024.42576639.parquet
|
| 46 |
+
- seismic_w_noise_vol_42576639.parquet
|
| 47 |
+
- ...
|
| 48 |
+
|
| 49 |
+
- test (15 noisy volumes)
|
| 50 |
+
- seismicCubes_RFC_fullstack_2023_1234567.parquet
|
| 51 |
+
- seismicCubes_RFC_fullstack_2023_1234568.parquet
|
| 52 |
+
- ...
|
| 53 |
+
- seismicCubes_RFC_fullstack_2023_1234568.parquet
|
| 54 |
+
|
| 55 |
+
## Dataset Creation
|
| 56 |
+
|
| 57 |
+
### Source Data
|
| 58 |
+
|
| 59 |
+
This data was produced by ThinkOnward for the Image Impeccable Challenge, using a synthetic seismic dataset generator called [Synthoseis](https://github.com/sede-open/synthoseis).
|
| 60 |
+
|
| 61 |
+
#### Who are the source data producers?
|
| 62 |
+
|
| 63 |
+
The data is provided as paired noisy and clean seismic volumes.
|
| 64 |
+
|
| 65 |
+
### Recommendations
|
| 66 |
+
|
| 67 |
+
This is a synthetically generated dataset, and differs from real-world seismic data. It is recommended that this dataset be used for research purposes only.
|
| 68 |
+
|
| 69 |
+
## Citation
|
| 70 |
+
|
| 71 |
+
**BibTeX:**
|
| 72 |
+
|
| 73 |
+
@misc {thinkonward_2025,
|
| 74 |
+
author = { {ThinkOnward} },
|
| 75 |
+
title = { },
|
| 76 |
+
year = 2025,
|
| 77 |
+
url = { https://huggingface.co/datasets/thinkonward/image-impeccable },
|
| 78 |
+
doi = { },
|
| 79 |
+
publisher = { Hugging Face }
|
| 80 |
+
}
|
| 81 |
+
|
| 82 |
+
## Dataset Card Contact
|
| 83 |
+
|
| 84 |
+
Please contact `challenges@thinkonward.com` for questions, comments, or concerns about this model.
|