TrafficMate is innovatively built upon a Multimodal framework, integrating diverse data sources to enhance traffic management and congestion prediction.
-
timeseries2raster: the algorithm to convert sensing data stored in time-series format to raster images.
-
prep3Draster: the algorithm to individual raster images to video-like data.
-
learning_model:
- 3D_CNN: a deep learning model predict information by preserving both spartial and temporal relationships using 3D-CNN.
- train_singlemodal.py: to train the model using only 1 factor (target factor)
- predict_singlemodal.py: to predict data using the model produced from train_singlemodal
- train_multimodal.py: to train the model using many factors predict_multimodal.py: to predict data using the model produced from train_multimodal
-
jpmesh: convert mesh code to coordinate
-
data_sample: sample dataset to run the algorithm