Paper: https://arxiv.org/abs/2308.14250
Dataset can be download here
This is the repo for paper Rule-Based Error Detection and Correction to Operationalize Movement Trajectory Classification
LabelMatrix-TimeDays-TrajectrotyMatrix.py data process step 1
Instance_Creation_sequence1.py data process step 2
special_LRCN_data.py generate train, valid, test data
train_predict.py train and test the model
speed_rule.py speed rule
rule_out.py improve the model performance through rule.
If you only want to use the rule_out.py file, just run this command:
python rule_out.py LRCN_F1_no_overlap_sequential no_overlap_sequential_10
LRCN_F1_no_overlap_sequential is the results(labels) from neural network
no_overlap_sequential_10 is the rule These two folders have more files than "python rule_out.py" needs, so just use the files that are needed.
Bowen Xi - [email protected] Paulo Shakarian - [email protected]