##DeepKnowledge @Designtime & @Runtime version
parent
├── DeepKnw_pack ← downloads here (1 MB)
└── Metadata
└── config.py ← data & model paths here
└── dataset ← download data here
└── data ← Designtime execution (needed for runtime)
└── experiments ← Designtime execution
└── Networks
└── KIOS ← trained model weights here
Each use case, i.e., KIOS, PAL, has its customized settings and metadata. These can be provided up to request.
cd DIRECTORY/DeepKnw_pack/
python setup.py install
####Example usage for DesignTime Deployment:
import DeepKnw_run as knw
COV=knw.DeepKnw(PATH_To_CONFIG_FILE)
test_path=PATH_To_TestData_Images
size=5000 (← EXPL)
test_loader = COV.getTestloader(test_path,size)
RSLT=COV.estimate_coverage(test_loader)
####Example usage for RunTime Deployment:
import DeepKnw_run as knw
COV=knw.DeepKnw(PATH_To_CONFIG_FILE)
YOLOloader, model_features=COV.DesignDataAnalyzer()
Frames_path=PATH_To_RunTime_Data
Deepknowledge_Uncertainty=COV.Runtime_Estimate(Frame_path,YOLOloader)