README.md

Created to address the issue of pothole problems on our way to work and especially back home.

POTHOLE DETECTION FRONTEND

Objctives: We have two frontends done by two devs: Hritik and Sakshi

◘ Frontend 1: set up an interface such that given a gps co-ordinate you must point it on map

◘ Frontend 2: set up an app interface such that upon an input of an image, the app will use our object detection model to detect the potholes in it.

Hritik has agreed on frontend #1 while Sakshi has agreed on frontend #2

POTHOLE DETECTION BACKEND

We shall be using an object detection model (either by tensorflow.js or YOLO(YOLO is chosen one for the prototype model) or render manually usng r-fcn)

We need training data of 200 simple potholes which are annotated.I have an image dataset of at least 1200 proper pothole images which I will upload below

https://drive.google.com/open?id=1TeqOhl8XdiBdkJOr72Q6mMvQ_WG2IzV5

Take any 30 simple potholes from your assigned 317 and annotate them (just choose your simple pothole images for now and we'll annotate them together. upload them here once you select at least 30. you could go for more too. i'd appreciate that immensely.)
Mayank: First 317

Raghvesh: 318-634 https://drive.google.com/open?id=1UdDMH3qUL4Fyb8zrMU9HYxSAu386deeK

prabhu: 635 to 951 https://drive.google.com/open?id=1idqJIEUPh8UrH1HfWdldkF3-pIPy_YGm

Imma(HBD!):remaining 317 https://drive.google.com/open?id=1QISgcLTDrDYZJsKQuVLdryT7pBt9njh-

Note

The previous json files are for trial purposes only(since they classify images and not detect objects in them). Could be used by both devs

https://drive.google.com/open?id=1Jh5-VRNY_7NO_GhZjWePM-u_Ys7WWHIh

Links

https://teachablemachine.withgoogle.com/faq
https://towardsdatascience.com/yolo-you-only-look-once-real-time-object-detection-explained-492dc9230006
https://www.youtube.com/watch?v=oHg5SJYRHA0


And most importantly,
https://www.pyimagesearch.com/2018/03/12/python-argparse-command-line-arguments/
then
https://www.pyimagesearch.com/2018/11/12/yolo-object-detection-with-opencv/

https://github.com/tzutalin/labelImg

https://github.com/thtrieu/darkflow

youtube tutorials

https://www.youtube.com/watch?v=Lg4T9iJkwhE (add more links if required)

https://drive.google.com/file/d/1gdu2YOqRE5Z4ayQ4-mQGpXN5LSv4qF-L/view?usp=sharing

Final Output:

Video Output