AGarioud commited on
Commit
5db24d4
1 Parent(s): d8f81c7

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +20 -19
README.md CHANGED
@@ -102,7 +102,7 @@ pipeline_tag: image-segmentation
102
  - **Compute infrastructure:**
103
  - software: python, pytorch-lightning
104
  - hardware: HPC/AI resources provided by GENCI-IDRIS
105
- - **License:** : Apache 2.0
106
 
107
  ---
108
 
@@ -184,24 +184,25 @@ Statistics of the TRAIN+VALIDATION set :
184
 
185
  #### Training Hyperparameters
186
 
187
- * Model architecture: Unet (implementation from the [Segmentation Models Pytorch library](https://segmentation-modelspytorch.readthedocs.io/en/latest/docs/api.html#unet))
188
- * Encoder : Resnet-34 pre-trained with ImageNet
189
- * Augmentation :
190
- * VerticalFlip(p=0.5)
191
- * HorizontalFlip(p=0.5)
192
- * RandomRotate90(p=0.5)
193
- * Input normalization (mean=0 | std=1):
194
- * norm_means: [105.08, 110.87, 101.82]
195
- * norm_stds: [52.17, 45.38, 44]
196
- * Seed: 2022
197
- * Batch size: 10
198
- * Number of epochs : 200
199
- * Early stopping : patience 30 and val_loss as monitor criterium
200
- * Optimizer : SGD
201
- * Schaeduler : mode = "min", factor = 0.5, patience = 10, cooldown = 4, min_lr = 1e-7
202
- * Learning rate : 0.02
203
- * Class Weights : [1-building: 1.0 , 2-pervious surface: 1.0 , 3-impervious surface: 1.0 , 4-bare soil: 1.0 , 5-water: 1.0 , 6-coniferous: 1.0 , 7-deciduous: 1.0 , 8-brushwood: 1.0 , 9-vineyard: 1.0 , 10-herbaceous vegetation: 1.0 , 11-agricultural land: 1.0 , 12-plowed land: 1.0 , 13-swimming_pool: 1.0 , 14-snow: 1.0 , 15-clear cut: 0.0 , 16-mixed: 0.0 , 17-ligneous: 0.0 , 18-greenhouse: 1.0 , 19-other: 0.0]
204
-
 
205
 
206
  #### Speeds, Sizes, Times
207
 
 
102
  - **Compute infrastructure:**
103
  - software: python, pytorch-lightning
104
  - hardware: HPC/AI resources provided by GENCI-IDRIS
105
+ - **License:** : Etalab 2.0
106
 
107
  ---
108
 
 
184
 
185
  #### Training Hyperparameters
186
 
187
+ ```yaml
188
+ - Model architecture: Unet #(implementation from the [Segmentation Models Pytorch library](https://segmentation-modelspytorch.readthedocs.io/en/latest/docs/api.html#unet))
189
+ - Encoder : Resnet-34 pre-trained with ImageNet
190
+ - Augmentation :
191
+ - VerticalFlip(p=0.5)
192
+ - HorizontalFlip(p=0.5)
193
+ - RandomRotate90(p=0.5)
194
+ - Input normalization (mean=0 | std=1):
195
+ - norm_means: [105.08, 110.87, 101.82]
196
+ - norm_stds: [52.17, 45.38, 44]
197
+ - Seed: 2022
198
+ - Batch size: 10
199
+ - Number of epochs : 200
200
+ - Early stopping : patience 30 and val_loss as monitor criterium
201
+ - Optimizer : SGD
202
+ - Schaeduler : mode = "min", factor = 0.5, patience = 10, cooldown = 4, min_lr = 1e-7
203
+ - Learning rate : 0.02
204
+ - Class Weights : [1-building: 1.0 , 2-pervious surface: 1.0 , 3-impervious surface: 1.0 , 4-bare soil: 1.0 , 5-water: 1.0 , 6-coniferous: 1.0 , 7-deciduous: 1.0 , 8-brushwood: 1.0 , 9-vineyard: 1.0 , 10-herbaceous vegetation: 1.0 , 11-agricultural land: 1.0 , 12-plowed land: 1.0 , 13-swimming_pool: 1.0 , 14-snow: 1.0 , 15-clear cut: 0.0 , 16-mixed: 0.0 , 17-ligneous: 0.0 , 18-greenhouse: 1.0 , 19-other: 0.0]
205
+ ```
206
 
207
  #### Speeds, Sizes, Times
208