Classification without Twist vs with Twist
Study without Twist
Learning Curves
Epoch 300/300
loss: 0.2271 - val_loss: 0.2393
accuracy: 0.9925 - val_accuracy: 0.9826
Using 937 samples for training and 401 for validation
Confusion Matrix
total test samples: 575
Correctly predicted: 566
Outputs Confidence Analysis
Legend:
- Max confidence reached (green line)
- Mean of every confidences (bold value)
- Standard deviation (right below mean)
- Min confidence reached (red line)
Wrongly Predicted Sample
Real action: Push
Predicted action: Shake
Outputs Confidence: [pull= 0.011 ; push= 0.429 ; shake= 0.5606]
Study with Twist
Learning Curves
Epoch 239/300
loss: 0.4024 - val_loss: 0.4805
accuracy: 0.9792 - val_accuracy: 0.9516
Using 1253 samples for training and 537 for validation
Using 1253 samples for training and 537 for validation
Confusion Matrix
total test samples: 768
Correctly predicted: 740
Outputs Confidence Analysis
Legend:
- Max confidence reached (green line)
- Mean of every confidences (bold value)
- Standard deviation (right below mean)
- Min confidence reached (red line)
Wrongly Predicted Sample
Real action: Shake
Predicted action: Twist
Outputs Confidence: [pull= 0.001 ; push= 0.011 ; shake= 0.001 ; twist= 0.988]
Comments
Post a Comment