Datasets:
Update README.md
Browse files
README.md
CHANGED
|
@@ -47,7 +47,6 @@ Color Fundus-to-RNFLT Norm predicts a personalized expected-normal retinal nerve
|
|
| 47 |
| File | Description |
|
| 48 |
| --- | --- |
|
| 49 |
| `checkpoint_best.pt` | Pretrained color fundus → personalized RNFLT norm model (~364 MB) |
|
| 50 |
-
| `mean_rnflt.npz` | Training-set mean RNFLT map used by the residual predictor (~140 KB) |
|
| 51 |
|
| 52 |
Both files are required. The checkpoint stores the exponential-moving-average weights.
|
| 53 |
|
|
@@ -59,7 +58,6 @@ Open **Files and versions** and download:
|
|
| 59 |
|
| 60 |
```text
|
| 61 |
checkpoint_best.pt
|
| 62 |
-
mean_rnflt.npz
|
| 63 |
```
|
| 64 |
|
| 65 |
Place the downloaded files in the `checkpoint/` directory of the color-fundus code repository:
|
|
@@ -116,8 +114,7 @@ from rnfltnorm.infer import (
|
|
| 116 |
rgb = load_fundus_rgb("image/fundus_1496028_1.png")
|
| 117 |
|
| 118 |
model, cfg, mean_np, device = load_predictor(
|
| 119 |
-
"checkpoint/checkpoint_best.pt"
|
| 120 |
-
"checkpoint/mean_rnflt.npz",
|
| 121 |
)
|
| 122 |
|
| 123 |
fundus = prepare_fundus(rgb, cfg, flip=False, crop_retina=False)
|
|
|
|
| 47 |
| File | Description |
|
| 48 |
| --- | --- |
|
| 49 |
| `checkpoint_best.pt` | Pretrained color fundus → personalized RNFLT norm model (~364 MB) |
|
|
|
|
| 50 |
|
| 51 |
Both files are required. The checkpoint stores the exponential-moving-average weights.
|
| 52 |
|
|
|
|
| 58 |
|
| 59 |
```text
|
| 60 |
checkpoint_best.pt
|
|
|
|
| 61 |
```
|
| 62 |
|
| 63 |
Place the downloaded files in the `checkpoint/` directory of the color-fundus code repository:
|
|
|
|
| 114 |
rgb = load_fundus_rgb("image/fundus_1496028_1.png")
|
| 115 |
|
| 116 |
model, cfg, mean_np, device = load_predictor(
|
| 117 |
+
"checkpoint/checkpoint_best.pt"
|
|
|
|
| 118 |
)
|
| 119 |
|
| 120 |
fundus = prepare_fundus(rgb, cfg, flip=False, crop_retina=False)
|