sezan92/carnd-advanced-lane-lines

Advanced Lane Detection

Advanced Lane Finding

HOW TO USE

1. Camera Calibration

python src/camera.py --cal_img_dir /path/to/calibration/chessboard/images --calibration_output_dir /path/to/directory/to/store/calibrated/images

more arguments

--input_img, input image to undistort
--output_dir, directory to save undistorted image
--cfg_filename, camera configuration file name

2. Perspective Transform

After calibrating the camera, we need to get the perspective transform of the road. This script helps us getting the transformation , and inverse transformation matrix.

python src/perspective_transform.py --image /path/to/image --cfg_filename /path/to/configuratio/file/to/save/configuration --camera_cfg_filename /path/to/camera/configuration/file

more arguments:

--threshold, threshold for perspective transformation. default: 400
--plot, to plot or not to plot.

3. Binary Transform

After transforming the perspective, we have to get tune the binary thresholds. So that we can get the binary images.

python  src/binary_transform.py --image_dir /path/to/images/to/get/binary/thresholds --cfg_filename /path/to/binary/threshold/configuration/file --camera_cfg_file /path/to/camera/configuration/file --pt_cfg_filename /path/to/perspective/transform/configuration/file 

more argument:

--img_size, image size, default: 720,1280

4. Lane Detection

After getting all the configuration files, i.e. Camera Calibration, Perspective Transformation, and Binary transformation , it is time to detect the lane.

From Directory of images

python src/lane_detection.py --input test_images --camera_cfg_filename config/camera_config.yaml --pt_cfg_filename config/pt_config.yaml --binary_cfg_filename config/camera_config.yaml --img_size 720 1280

It will save the output images in test_images/lane_detected folder

From Video

python src/lane_detection.py --input project_video.mp4 --camera_cfg_filename config/camera_config.yaml --pt_cfg_filename config/pt_config.yaml --binary_cfg_filename config/camera_config.yaml --img_size 720 1280

Video Link

It will save the output video as project_video_output.avi .

Notebook

I have also written Notebook on this project. That is available here

TODO:

• Tune more to get better results
• Work on Challenge Videos
• Make a Docker environment
• Refactor
• Make a package
• Refactor test scripts

carnd-advanced-lane-lines's Issues

TODO

• Sobel Y
• Combine Gradients

return matrix or corners?

I am not sure if i will return camera matrix or corner in get_calibration_matrix?