AGarioud commited on
Commit
bbee304
·
verified ·
1 Parent(s): 4eacf82

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +7 -11
README.md CHANGED
@@ -6,21 +6,17 @@ tags:
6
  - pytorch
7
  - landcover
8
  library_name: pytorch
 
 
9
  ---
10
 
11
- We introduce MAESTRO, a tailored adaptation of the Masked Autoencoder (MAE) framework that effectively orchestrates the use of multimodal, multitemporal, and multispectral Earth Observation (EO) data. Evaluated on four EO datasets, MAESTRO sets a new state-of-the-art on tasks that strongly rely on multitemporal dynamics, while remaining highly competitive on tasks dominated by a single monotemporal modality.
12
 
13
  Our contributions are as follows:
14
-
15
- Extensive benchmarking of multimodal and multitemporal SSL: Impact evaluation of various fusion strategies for multimodal and multitemporal SSL.
16
- Patch-group-wise normalization: Novel normalization scheme that normalizes reconstruction targets patch-wise within groups of highly correlated spectral bands.
17
- MAESTRO: Novel adaptation of the MAE that combines optimized fusion strategies with our tailored patch-group-wise normalization.
18
 
19
  <div style="position: relative; text-align: center;">
20
  <img src="./media/Maestro_Overview.png" alt="Classes distribution." style="width: 100%; display: block; margin: 0 auto;"/>
21
- </div>
22
-
23
-
24
-
25
-
26
-
 
6
  - pytorch
7
  - landcover
8
  library_name: pytorch
9
+ datasets:
10
+ - IGNF/FLAIR-HUB
11
  ---
12
 
13
+ We introduce **MAESTRO**, a tailored adaptation of the Masked Autoencoder (MAE) framework that effectively orchestrates the use of multimodal, multitemporal, and multispectral Earth Observation (EO) data. Evaluated on four EO datasets, MAESTRO sets a new state-of-the-art on tasks that strongly rely on multitemporal dynamics, while remaining highly competitive on tasks dominated by a single monotemporal modality.
14
 
15
  Our contributions are as follows:
16
+ - **Extensive benchmarking of multimodal and multitemporal SSL:** Impact evaluation of various fusion strategies for multimodal and multitemporal SSL.
17
+ - **Patch-group-wise normalization:** Novel normalization scheme that normalizes reconstruction targets patch-wise within groups of highly correlated spectral bands.
18
+ - **MAESTRO:** Novel adaptation of the MAE that combines optimized fusion strategies with our tailored patch-group-wise normalization..
 
19
 
20
  <div style="position: relative; text-align: center;">
21
  <img src="./media/Maestro_Overview.png" alt="Classes distribution." style="width: 100%; display: block; margin: 0 auto;"/>
22
+ </div>