Hugging Face's logo

sugam24
/
geonusaf-unet-r18-block-fold1

segmentation-models-pytorch
semantic-segmentation
remote-sensing
unet
resnet18
geonusaf
Model card Files
xet
 Community
sugam24 commited on
Commit
4223513
·
verified ·
1 Parent(s): 0a101be

card epoch 62

Browse files
Files changed (1) hide show
  1. README.md +11 -11
README.md CHANGED
@@ -19,20 +19,20 @@ Weights are the **EMA** weights (decay 0.999), not the raw final weights.
19
  | input | 512x512, ImageNet norm, effective GSD 0.586 m/px |
20
  | regularization | wd 0.01 (norm/bias exempt), ls 0.05, drop 0.1, EMA 0.999 |
21
  | classes | Residential, Road, River, Forest, UnusedLand, Agricultural |
22
- | best epoch | 60 |
23
- | val mIoU | 0.4487 |
24
- | val mF1 | 0.5784 |
25
- | val OA | 0.7864 |
26
- | val kappa | 0.6201 |
27
 
28
  ## Per-class (validation)
29
  | class | IoU | F1 |
30
  |---|---|---|
31
- | Residential | 0.8294 | 0.9067 |
32
- | Road | 0.3835 | 0.5544 |
33
- | River | 0.1244 | 0.2212 |
34
- | Forest | 0.6465 | 0.7853 |
35
- | UnusedLand | 0.1904 | 0.3199 |
36
- | Agricultural | 0.5183 | 0.6828 |
37
 
38
  Checkpoint `best.pt` holds `model_state` (EMA) plus `cfg`, `metrics` and `arch_sig`.
 
19
  | input | 512x512, ImageNet norm, effective GSD 0.586 m/px |
20
  | regularization | wd 0.01 (norm/bias exempt), ls 0.05, drop 0.1, EMA 0.999 |
21
  | classes | Residential, Road, River, Forest, UnusedLand, Agricultural |
22
+ | best epoch | 62 |
23
+ | val mIoU | 0.4516 |
24
+ | val mF1 | 0.5807 |
25
+ | val OA | 0.7898 |
26
+ | val kappa | 0.6255 |
27
 
28
  ## Per-class (validation)
29
  | class | IoU | F1 |
30
  |---|---|---|
31
+ | Residential | 0.8310 | 0.9077 |
32
+ | Road | 0.3965 | 0.5678 |
33
+ | River | 0.1329 | 0.2346 |
34
+ | Forest | 0.6495 | 0.7875 |
35
+ | UnusedLand | 0.1763 | 0.2998 |
36
+ | Agricultural | 0.5232 | 0.6870 |
37
 
38
  Checkpoint `best.pt` holds `model_state` (EMA) plus `cfg`, `metrics` and `arch_sig`.