Human Activity Recognition

Paper : Paper_Human Activity Recognition

Results:

1. Data Preprocessing
2. Input Pipeline
3. Model architecture
4. Metrics
   • Confusion Matrix
   • ROC AUC Curve
5. Hyperparamter Optimization
   • Neurons in LSTM layer
   • Optimizer
   • Dropout rate

Results obtained so far:

Accuracy Test Data Set : 77.23 %

1. Training Data Visualization

2. Test Data Visualization

3. Posture Data Visualization

WALKING

WALKING_UPSTAIRS

WALKING_DOWNSTAIRS

SITTING

STANDING

LAYING

STAND_TO_SIT

SIT_TO_STAND

SIT_TO_LIE

LIE_TO_SIT

STAND_TO_LIE

LIE_TO_STAND

4. Histogram: Training Examples per Activity

5. Metrics

6. Confusion Matrix

7. ROC-AUC Curve