Lane Detection using Deep Learning

Group project for Deep Learning Class- CMPE 258

Group Members:-

Aditya Sahu
Divyam Sobti
Prerna Shekar Bharadwaj
Shashank Sharma

How to run the project

Library Requirements

Tensorflow 2.10.0
Python 3.7.16
opencv 4.6.0
numpy 1.21.5
keras 2.10.0
scikit-learn 1.0.2

All other requirements are mentioned in requirements.txt file

Project Code Files Description:

In order to train the model, we have developed the code file “train_NN.py”. This file is responsible for training the model which is under the ‘training’ folder.

In order to run the model, we have developed the code file “detected_lanes_main.py”. This file takes input as a video file path and performs Lane Detection.

Steps for training the Model:-

i) Download the dataset, full_CNN_train.p and full_CNN_labels.p files from the below google drive link:- https://drive.google.com/drive/u/0/folders/1DlATz2oC-ui72HN66YDtkUvORjlDgL3f

ii) Execute the code through the below command.

python  train_NN.py

iii) Once the code runs, the model is saved as a CNN_model.h5 file. This model is used for detecting the lanes in detected_lanes_main.py file.

Steps for executing the Code:

i) For executing the program, pass “video_path” as parameter during running the program using the following command:-

python  detected_lanes_main.py --input video_path 

ii) Once the program is run, the file is saved as Output1.mp4