sgiordano commited on
Commit
d0120df
1 Parent(s): 37c1545

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +18 -18
README.md CHANGED
@@ -212,10 +212,10 @@ Statistics of the TRAIN+VALIDATION set :
212
 
213
  #### Speeds, Sizes, Times
214
 
215
- The FLAIR-INC_rgbie_15cl_resnet34-unet model was trained on a HPC/AI resources provided by GENCI-IDRIS (Grant 2022-A0131013803).
216
  16 V100 GPUs were used ( 4 nodes, 4 GPUS per node). With this configuration the approximate learning time is 6 minutes per epoch.
217
 
218
- FLAIR-INC_rgbie_15cl_resnet34-unet was obtained for num_epoch=76 with corresponding val_loss=0.56.
219
 
220
 
221
  <div style="position: relative; text-align: center;">
@@ -242,29 +242,29 @@ As a result the _Snow_ class is absent from the TEST set.
242
 
243
  #### Metrics
244
 
245
- With the evaluation protocol, the **FLAIR-INC_RVBIE_resnet34_unet_15cl_norm** have been evaluated to **OA= 76.37%** and **mIoU=58.63%**.
246
  The _snow_ class is discarded from the average metrics.
247
 
248
  The following table give the class-wise metrics :
249
 
250
  | Modalities | IoU (%) | Fscore (%) | Precision (%) | Recall (%) |
251
  | ----------------------- | ----------|---------|---------|---------|
252
- | building | 82.63 | 90.49 | 90.26 | 90.72 |
253
- | pervious surface | 53.24 | 69.48 | 68.97 | 70.00 |
254
- | impervious surface | 74.17 | 85.17 | 86.28 | 84.09 |
255
- | bare soil | 60.40 | 75.31 | 80.49 | 70.75 |
256
- | water | 87.59 | 93.38 | 93.16 | 93.61 |
257
- | coniferous | 46.35 | 63.34 | 63.52 | 63.16 |
258
- | deciduous | 67.45 | 80.56 | 77.44 | 83.94 |
259
- | brushwood | 30.23 | 46.43 | 63.55 | 36.58 |
260
- | vineyard | 82.93 | 90.67 | 91.35 | 89.99 |
261
- | herbaceous vegetation | 55.03 | 70.99 | 70.59 | 71.40 |
262
- | agricultural land | 52.01 | 68.43 | 59.18 | 81.12 |
263
- | plowed land | 40.84 | 57.99 | 68.28 | 50.40 |
264
- | swimming_pool | 48.44 | 65.27 | 81.62 | 54.37 |
265
  | _snow_ | _00.00_ | _00.00_ | _00.00_ | _00.00_ |
266
- | greenhouse | 39.45 | 56.57 | 45.52 | 74.72 |
267
- | **average** | **58.63** | **72.44** | **74.3** | **72.49** |
268
 
269
 
270
 
 
212
 
213
  #### Speeds, Sizes, Times
214
 
215
+ The FLAIR-INC_rgbi_15cl_resnet34-unet model was trained on a HPC/AI resources provided by GENCI-IDRIS (Grant 2022-A0131013803).
216
  16 V100 GPUs were used ( 4 nodes, 4 GPUS per node). With this configuration the approximate learning time is 6 minutes per epoch.
217
 
218
+ FLAIR-INC_rgbi_15cl_resnet34-unet was obtained for num_epoch=65 with corresponding val_loss=0.56.
219
 
220
 
221
  <div style="position: relative; text-align: center;">
 
242
 
243
  #### Metrics
244
 
245
+ With the evaluation protocol, the **FLAIR-INC_rgbi_15cl_resnet34-unet** have been evaluated to **OA= 76.26%** and **mIoU=56.67%**.
246
  The _snow_ class is discarded from the average metrics.
247
 
248
  The following table give the class-wise metrics :
249
 
250
  | Modalities | IoU (%) | Fscore (%) | Precision (%) | Recall (%) |
251
  | ----------------------- | ----------|---------|---------|---------|
252
+ | building | 78.61 | 88.03 | 88.60 | 87.47 |
253
+ | pervious surface | 52.66 | 68.99 | 70.90 | 67.18 |
254
+ | impervious surface | 72.57 | 84.10 | 84.00 | 84.20 |
255
+ | bare soil | 59.66 | 74.73 | 77.46 | 72.18 |
256
+ | water | 87.6 | 93.39 | 92.37 | 94.44 |
257
+ | coniferous | 62.25 | 76.74 | 77.34 | 76.14 |
258
+ | deciduous | 71.76 | 83.56 | 81.67 | 85.54 |
259
+ | brushwood | 31.27 | 47.64 | 59.51 | 39.72 |
260
+ | vineyard | 76.45 | 86.66 | 85.32 | 88.03 |
261
+ | herbaceous vegetation | 51.64 | 68.11 | 70.79 | 65.63 |
262
+ | agricultural land | 57.64 | 73.13 | 66.93 | 80.59 |
263
+ | plowed land | 43.45 | 60.58 | 58.84 | 62.43 |
264
+ | swimming_pool | 41.26 | 58.42 | 79.5 | 46.17 |
265
  | _snow_ | _00.00_ | _00.00_ | _00.00_ | _00.00_ |
266
+ | greenhouse | 63.22 | 77.47 | 69.99 | 86.73 |
267
+ | **average** | **56.67** | **69.44** | **70.88** | **69.1** |
268
 
269
 
270