Commit
·
72384c6
1
Parent(s):
b1d35fb
Update README.md
Browse files
README.md
CHANGED
@@ -4,10 +4,10 @@ library_name: keras
|
|
4 |
|
5 |
## Model description
|
6 |
|
7 |
-
Autoencoder model trained to compress information from sentinel-2 satellite images using Resnet50 V2 as
|
8 |
The latent space of the model is given by 1024 neurons which can be used to generate embeddings from the sentinel-2 satellite images.
|
9 |
|
10 |
-
The model was trained using bands 2, 3 and 4 of the Sentinel-2 satellites and
|
11 |
|
12 |
The input shape of the model is 224, 224, 3. To extract features you should remove the last layer.
|
13 |
|
@@ -17,8 +17,9 @@ The model was trained with images of 81 different cities in Colombia, however it
|
|
17 |
|
18 |
## Training and evaluation data
|
19 |
|
20 |
-
The model was trained with satellite images of 81 different cities in Colombia extracted from sentinel-2.
|
21 |
|
|
|
22 |
## Training procedure
|
23 |
|
24 |
### Training hyperparameters
|
|
|
4 |
|
5 |
## Model description
|
6 |
|
7 |
+
Autoencoder model trained to compress information from sentinel-2 satellite images using Resnet50 V2 as encoder backbone to extract features.
|
8 |
The latent space of the model is given by 1024 neurons which can be used to generate embeddings from the sentinel-2 satellite images.
|
9 |
|
10 |
+
The model was trained using bands RGB (2, 3 and 4) (Red, Green and Blue) of the Sentinel-2 satellites and using 81 municipalities of Colombia with most dengue cases.
|
11 |
|
12 |
The input shape of the model is 224, 224, 3. To extract features you should remove the last layer.
|
13 |
|
|
|
17 |
|
18 |
## Training and evaluation data
|
19 |
|
20 |
+
The model was trained with satellite images of 81 different cities in Colombia extracted from sentinel-2 using RGB bands using an asymmetric autoencoder. Images with information that could result in noise such as black images were filtered prior to training to avoid noise in the data.
|
21 |
|
22 |
+
The dataset was split into train and test using 80% for train and 20% to test.
|
23 |
## Training procedure
|
24 |
|
25 |
### Training hyperparameters
|