Suggestion From ChenRui

[developerChenRui]

After inspecting of your project, I have several suggestions which I hope will be helpful for your project.

When I tried to run the line_fit_video, the IDE reminds me of lacking Line. Then I pip install Line. But the error still exist. Later I find another Line.py in another file folder. What I want to say is that managing your GitHub is necessary!!! Please update the readme and specify the steps to run the project, put all necessary codes into one file so that people can run your project directly!!!

1. I notice in the line_fit_video.py, there are some imported packages but not used. And the style of python code is not standard. Use the method(pep8 pylint) we learn from 602 !!
2. There are so many files in your project contributing to that I do not know exactly what I should focus on. So my suggestion is to remove the unnecessary ones and just keep the useful files!!
3. The environment needed for running you project is so limited. I mean maybe making it fit for as many as possible environments is necessary.
4. Developing a mobile app (or something people can directly use) with your project is necessary because not everyone who wants to use your project is programmer who knows how to set up environment.
5. I compare the original project and yours. I find that it is not so clear that what you have improved for the original project. I hope you can display it in the Readme such as the accuracy, the model, any improvements. Or maybe you have found many algorithms but have not yet integrated them together ? I am looking forward to seeing them in the README.
6. I have noticed the challenges (Bird’s Eye View transform/Accuracy/environments) you mentioned in the sprint3, how did you deal with those challenges?Have you solved those problems?
7. I notice all methods the project use are related to cv. Yes, cv is perfect for video tracking, but diving into the technology behind (like machine learning, training the model by yourself ) I think will also be exciting.

Video tracking is an amazing issue related to the machine learning and image processing and a very hot issue--Self-Driving Cars. I will keep following your excellent project!!
For better knowing about your project, I also collect something useful about track detection. I would like to share them here:
Basic steps for track detection
Gray Scale Transformation
Gaussian Smoothing
Canny Edge Detection
ROI (Region of Interest) Based Edge Filtering
Hough Transformation
Lane Extrapolation

High ISO Noise Reduction: http://www.learn.usa.canon.com/resources/articles/2011/high_iso_noise_reduction_article.shtml

Gaussian Smoothing
http://homepages.inf.ed.ac.uk/rbf/HIPR2/gsmooth.htm

GitHub : https://github.com/feixia586/zhihu_material/tree/master/car_lane_detection

Let me know if it helps!