Project by Onkar Trivedi, Aishwary Jagetia & Animesh Nema
Dataset collected and made available by MIT AgeLab. Acquired from the following link: http://lexfridman.com/carsync/
The CNN can be run by the following steps:
Parsing the main datset: Main_Dataset_Parsing.py (converts the main video_front file into individual frames)
Generating a corresponding segmented dataset: Road_segmentation.py
Running the final CNN model: DLmodel_final_CNN.py
Visualizing the CNN model: CNNmodel_Visualization.py
Generating a video to visualize the model results: Slope_Visualization_video.py https://www.dropbox.com/sh/mjra6t2qfae8s56/AACC5cWL4Zs0DCpZabBjdAida?dl=0
[NOTE: The model.h5 & model.json files are provided, generated from the final CNN model. road_dataset_newest.csv & roadseg_dataset_newest.csv contain the respective csv files with the required IMU pitch & GPS altitude data.]