Instructions to use abdourahmane01/noise2void with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Keras
How to use abdourahmane01/noise2void with Keras:
# Available backend options are: "jax", "torch", "tensorflow". import os os.environ["KERAS_BACKEND"] = "jax" import keras model = keras.saving.load_model("hf://abdourahmane01/noise2void") - Notebooks
- Google Colab
- Kaggle
Upload folder using huggingface_hub
Browse files- README.md +21 -0
- config.json +1 -0
- weights_best.h5 +3 -0
- weights_last.h5 +3 -0
README.md
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Noise2Void
|
| 2 |
+
|
| 3 |
+
Modèle de débruitage d'images par auto-supervision (self-supervised denoising), entraîné avec la librairie [N2V](https://github.com/juglab/n2v) (TensorFlow/Keras).
|
| 4 |
+
|
| 5 |
+
## Fichiers
|
| 6 |
+
- `weights_best.h5` : meilleurs poids (selon validation loss)
|
| 7 |
+
- `weights_last.h5` : derniers poids de l'entraînement
|
| 8 |
+
- `config.json` : configuration de l'entraînement N2V
|
| 9 |
+
|
| 10 |
+
## Utilisation
|
| 11 |
+
|
| 12 |
+
```python
|
| 13 |
+
from n2v.models import N2V, N2VConfig
|
| 14 |
+
|
| 15 |
+
config = N2VConfig(...) # ou charger depuis config.json
|
| 16 |
+
model = N2V(config=config, name='n2v_model_test_1_data_gray', basedir='.')
|
| 17 |
+
model.keras_model.load_weights('weights_best.h5')
|
| 18 |
+
```
|
| 19 |
+
|
| 20 |
+
## Reproductibilité
|
| 21 |
+
Ce modèle a été entraîné dans le cadre d'un projet de recherche sur le débruitage d'images médicales. Voir `config.json` pour les hyperparamètres exacts utilisés lors de l'entraînement.
|
config.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"means": ["0.5447642"], "stds": ["0.118812814"], "n_dim": 2, "axes": "YXC", "n_channel_in": 1, "n_channel_out": 1, "unet_residual": false, "unet_n_depth": 3, "unet_kern_size": 3, "unet_n_first": 64, "unet_last_activation": "linear", "unet_input_shape": [null, null, 1], "train_loss": "mse", "train_epochs": 100, "train_steps_per_epoch": 1142, "train_learning_rate": 0.0004, "train_batch_size": 32, "train_tensorboard": true, "train_checkpoint": "weights_best.h5", "train_reduce_lr": {"factor": 0.5, "patience": 10}, "batch_norm": true, "n2v_perc_pix": 0.198, "n2v_patch_shape": [64, 64], "n2v_manipulator": "uniform_withCP", "n2v_neighborhood_radius": 5, "single_net_per_channel": true, "blurpool": false, "skip_skipone": false, "structN2Vmask": null, "probabilistic": false}
|
weights_best.h5
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6d49c4f18c98c152f9593af10d90a9859e34589afc9fad495d4fcfcb1d76159e
|
| 3 |
+
size 16833360
|
weights_last.h5
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b38f80dd1cd509320a82f424e36a9f80d04e56bf6a56ed8e17322f1feb8e5c63
|
| 3 |
+
size 16833360
|