Datasets:
Tasks:
Image Segmentation
Modalities:
Image
Sub-tasks:
semantic-segmentation
Languages:
English
Size:
10K - 100K
License:
Update README.md
Browse files
README.md
CHANGED
@@ -21,3 +21,95 @@ task_categories:
|
|
21 |
task_ids:
|
22 |
- semantic-segmentation
|
23 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
21 |
task_ids:
|
22 |
- semantic-segmentation
|
23 |
---
|
24 |
+
|
25 |
+
|
26 |
+
## Setup
|
27 |
+
|
28 |
+
```bash
|
29 |
+
pip install datasets
|
30 |
+
```
|
31 |
+
|
32 |
+
## Feel the Magic
|
33 |
+
|
34 |
+
### Load Dataset
|
35 |
+
```python
|
36 |
+
from datasets import load_dataset
|
37 |
+
|
38 |
+
data = load_dataset('ziq/RSNA-ATD2023')
|
39 |
+
print(data)
|
40 |
+
```
|
41 |
+
|
42 |
+
```bash
|
43 |
+
DatasetDict({
|
44 |
+
train: Dataset({
|
45 |
+
features: ['patient_id', 'series_id', 'frame_id', 'image', 'mask'],
|
46 |
+
num_rows: 70291
|
47 |
+
})
|
48 |
+
})
|
49 |
+
```
|
50 |
+
|
51 |
+
### Train Test Split
|
52 |
+
```python
|
53 |
+
data = data['train'].train_test_split(test_size=0.2)
|
54 |
+
```
|
55 |
+
|
56 |
+
```python
|
57 |
+
train, test = data['train'], data['test']
|
58 |
+
|
59 |
+
# train[0]['patient_id']
|
60 |
+
# train[0]['image'] -> PIL Image
|
61 |
+
# train[0]['mask'] -> PIL Image
|
62 |
+
```
|
63 |
+
|
64 |
+
### Get Segmentation Mask
|
65 |
+
|
66 |
+
```python
|
67 |
+
ids = 3
|
68 |
+
image, mask = train[ids]['image'], train[ids]['mask']
|
69 |
+
|
70 |
+
mask = np.array(mask)
|
71 |
+
liver, spleen, right_kidney, left_kidney, bowel = [(mask == i,1,0)[0] * i for i in range(1, len(labels))]
|
72 |
+
```
|
73 |
+
|
74 |
+
|
75 |
+
### Plot Mask
|
76 |
+
|
77 |
+
```python
|
78 |
+
from matplotlib.colors import ListedColormap, BoundaryNorm
|
79 |
+
|
80 |
+
colors = ['#02020e', '#520e6d', '#c13a50', '#f57d15', '#fac62c', '#f4f88e'] # inferno
|
81 |
+
bounds = range(0, len(colors) + 1)
|
82 |
+
|
83 |
+
# Define the boundaries for each class in the colormap
|
84 |
+
cmap, norm = ListedColormap(colors), BoundaryNorm(bounds, len(colors))
|
85 |
+
|
86 |
+
# Plot the segmentation mask with the custom colormap
|
87 |
+
def plot_mask(mask, alpha=1.0):
|
88 |
+
_, ax = plt.subplots()
|
89 |
+
cax = ax.imshow(mask, cmap=cmap, norm=norm, alpha=alpha)
|
90 |
+
cbar = plt.colorbar(cax, cmap=cmap, norm=norm, boundaries=bounds, ticks=bounds)
|
91 |
+
cbar.set_ticks([])
|
92 |
+
_labels = [""] + labels
|
93 |
+
for i in range(1, len(_labels)):
|
94 |
+
cbar.ax.text(2, -0.5 + i, _labels[i], ha='left', color=colors[i - 1], fontsize=8)
|
95 |
+
plt.axis('off')
|
96 |
+
plt.show()
|
97 |
+
|
98 |
+
def plot(image, mask, cmap='gray'):
|
99 |
+
plt.imshow(image * np.where(mask > 0,1,0), cmap=cmap)
|
100 |
+
plt.axis('off')
|
101 |
+
plt.show()
|
102 |
+
```
|
103 |
+
|
104 |
+
```python
|
105 |
+
plot(image, mask)
|
106 |
+
```
|
107 |
+
|
108 |
+
![gray cmap](https://huggingface.co/datasets/ziq/RSNA-ATD2023/resolve/main/assets/grayscale.png)
|
109 |
+
|
110 |
+
```python
|
111 |
+
plot_mask(mask)
|
112 |
+
```
|
113 |
+
|
114 |
+
![custom cmap](https://huggingface.co/datasets/ziq/RSNA-ATD2023/resolve/main/assets/mask.png)
|
115 |
+
|