Datasets:
Update Dataset Card
#1
by
tayden
- opened
README.md
CHANGED
@@ -1,3 +1,102 @@
|
|
1 |
-
---
|
2 |
-
license: cc-by-4.0
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: cc-by-4.0
|
3 |
+
task_categories:
|
4 |
+
- image-segmentation
|
5 |
+
language:
|
6 |
+
- en
|
7 |
+
tags:
|
8 |
+
- biology
|
9 |
+
- climate
|
10 |
+
pretty_name: MusselSeg
|
11 |
+
size_categories:
|
12 |
+
- 1K<n<10K
|
13 |
+
---
|
14 |
+
|
15 |
+
# MusselSeg: Semantic Segmentation Dataset for Rocky Intertidal Mussel Habitat
|
16 |
+
|
17 |
+
## Dataset description
|
18 |
+
|
19 |
+
MusselSeg is a large-scale dataset for semantic segmentation of mussel habitat using high resolution drone imagery. It covers coastal mussel habitat located on the central coast of British Columbia, Canada, as well as areas in California, USA and provides pixel-wise annotation for mussel beds.
|
20 |
+
|
21 |
+
- **Source:** Imagery collected by the Hakai Institute and University of California Santa Cruz
|
22 |
+
- **Geographic Coverage:** ~3500m^2
|
23 |
+
- **Time Period:** 2021-2023
|
24 |
+
|
25 |
+
## Task description
|
26 |
+
|
27 |
+
The dataset is designed for semantic segmentation of mussel habitat in aerial imagery. The task involves assigning each pixel in the image to either the "mussel" class or "other"/"background" class.
|
28 |
+
|
29 |
+
## Data characteristics
|
30 |
+
|
31 |
+
- **Image Format:** GeoTiff
|
32 |
+
- **Resolution:** mean=0.45cm, stdev=0.20cm
|
33 |
+
- **Tile Size:** 1024x1024 pixels with 50% overlap (*TODO* check this)
|
34 |
+
- **Number of Tiles:** 9972 image and label pairs
|
35 |
+
- **Total Dataset Size:** 42G
|
36 |
+
## Annotation details
|
37 |
+
- **Method:** Manual heads-up digitizing with manual verification
|
38 |
+
- **Format:** Pixel-wise labels stored as separate mask images
|
39 |
+
- **Labelling Convention:** Each pixel assigned a single class label
|
40 |
+
|
41 |
+
## Class distribution
|
42 |
+
|
43 |
+
| Class ID | Class Name | Description | Percentage |
|
44 |
+
| -------- | ---------- | ------------------ | ---------- |
|
45 |
+
| 0 | Background | Unclassified areas | 87% |
|
46 |
+
| 1 | Mussels | Mussel bed | 13% |
|
47 |
+
|
48 |
+
## Split information
|
49 |
+
|
50 |
+
- Train: 48% (4834 tiles)
|
51 |
+
- Validation 13% (1277 tiles)
|
52 |
+
- Test 39% (3861 tiles)
|
53 |
+
|
54 |
+
Train and Validation split tiles all contain at least 1 pixel in each class. For the Test split, some tiles are entirely the background class.
|
55 |
+
|
56 |
+
Splits are created such that tiles from the same source orthomosaic image are not divided across different splits. That is, all tiles derived from the same drone flight are present only in a single split.
|
57 |
+
|
58 |
+
## Preprocessing
|
59 |
+
|
60 |
+
1. Orthorectification applied to raw imagery
|
61 |
+
2. Tiles extracted with 50% overlap
|
62 |
+
3. Tiles with no mussels present eliminated for the Train and Validation splits
|
63 |
+
|
64 |
+
## Licensing information
|
65 |
+
|
66 |
+
This dataset is released under the Creative Commons Attribution 4.0 License (CC BY 4.0).
|
67 |
+
|
68 |
+
## Ethical considerations
|
69 |
+
|
70 |
+
- No identifiable individuals are present in imagery
|
71 |
+
- Minimized impact on wildlife and sensitive habitats
|
72 |
+
- Engaged with local First Nations in planning aerial surveys
|
73 |
+
|
74 |
+
## Citation information
|
75 |
+
|
76 |
+
If you use this dataset in your research, please cite:
|
77 |
+
|
78 |
+
*TODO* Add authors? Update this
|
79 |
+
```
|
80 |
+
@dataset{MusselSeg2024,
|
81 |
+
author = {Denouden, T., McInnes, W.},
|
82 |
+
title = {MusselSeg: Semantic Segmentation Dataset for Rocky Intertidal Mussel Habitat},
|
83 |
+
year = {2024},
|
84 |
+
publisher = {Hugging Face},
|
85 |
+
howpublished = {\url{https://huggingface.co/datasets/HakaiInstitute/rgb-mussel-presence-1024-1024}}
|
86 |
+
}
|
87 |
+
```
|
88 |
+
|
89 |
+
## Baseline model performance
|
90 |
+
|
91 |
+
| Model | IoU for Mussel Class | Pixel Accuracy |
|
92 |
+
| ---------- | -------------------- | -------------- |
|
93 |
+
| U-Net | *TODO* | *TODO* |
|
94 |
+
| DeepLabV3+ | *TODO* | *TODO* |
|
95 |
+
|
96 |
+
## Known limitations
|
97 |
+
- Limited seasonal variation due to imagery being captured primarily in summer months
|
98 |
+
- Imagery only covers areas with known mussel beds
|
99 |
+
- Limited examples of mussel beds in urban or built-up environments
|
100 |
+
|
101 |
+
## Additional resources
|
102 |
+
- [GitHub Repository](https://github.com/HakaiInstitute/kelp-o-matic)
|