javier/Photovoltaic_Fault_Detector
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

.....

Browse Source
This commit is contained in:
Daniel Saavedra
2020-12-23 21:41:24 -03:00
parent 9f1566fef8
commit d5424bac82
18 changed files with 166 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
Result_yolo3_panel2/yolo3_full_yolo.output Normal file
View File

@@ -0,0 +1,22 @@
Seen labels: {'panel': 854}
Given labels: ['panel']
Training on: ['panel']
multi_gpu:1
Loading pretrained weights.
Train for 12 steps, validate for 1 steps
Epoch 1/2
Epoch 00001: loss improved from inf to 40.76296, saving model to Result_yolo3_panel2/yolo3_full_panel.h5
12/12 - 68s - loss: 40.7630 - yolo_layer_loss: 12.8884 - yolo_layer_1_loss: 27.8505 - yolo_layer_2_loss: 0.0241 - val_loss: 3793.7336 - val_yolo_layer_loss: 1322.7670 - val_yolo_layer_1_loss: 2470.9607 - val_yolo_layer_2_loss: 0.0060
Epoch 2/2
Epoch 00002: loss did not improve from 40.76296
12/12 - 71s - loss: 46.9910 - yolo_layer_loss: 20.9780 - yolo_layer_1_loss: 26.0097 - yolo_layer_2_loss: 0.0033 - val_loss: 72523.8125 - val_yolo_layer_loss: 4534.5332 - val_yolo_layer_1_loss: 67989.2812 - val_yolo_layer_2_loss: 0.0000e+00
94 instances of class panel with average precision: 0.0000
mAP using the weighted average of precisions among classes: 0.0000
mAP: 0.0000