shubham25kr07 / ml-pot-hole-detection

Intelligent Pot Hole detection using imaging and computer vision.

Intelligent Pothole Detection

Intelligent Pothole detection using imaging and computer vision.

Problem Statement

Bowl-shaped holes of various sizes in the pavement surface. Minimum plan dimension is 150 mm. Circular potholes should have a minimum diameter of 150 mm. A 150-mmdiameter circle should fit inside irregular-shaped potholes.

Literature Review

Generic

1. https://rosap.ntl.bts.gov/view/dot/28612
2. https://www.hindawi.com/journals/mpe/2015/968361/

2D image-based approach

1. https://repository.up.ac.za/bitstream/handle/2263/57733/Nienaber_Detecting_2015.pdf?sequence=1
2. http://www.icter.org/conference/icter2016/sites/default/files/icter/IITC-2008p1.pdf
3. https://pdfs.semanticscholar.org/f28c/b61a43f16bde57072819b107e0051c6c0afc.pdf
4. http://www.wseas.us/e-library/conferences/2013/Antalya/ITCN/ITCN-05.pdf
5. http://www.ijetsr.com/images/short_pdf/1491839718_dmce906_venue_ijetsr.pdf
6. http://cisjournal.org/journalofcomputing/archive/vol8no1/vol8no1_5.pdf

Vibration-based method

1. https://medium.com/@percepsense/intelligent-pothole-detection-879ef635dd38
2. https://github.com/marty-Wallace/Pothole

Solution Approach

Approach

Define the Technical Problem

1. Samples Collection http://www.image-net.org/synset?wnid=n09398076
2. Image Preprocessing
3. Visualization
4. Feature Selection
   • Color Model
   • Contour Detection
   • Shape Detection

• Contour drawing
• Perspective (Not included)
• Grayscale
• Color Tone

Training

1. Training Data Preparation
2. Algorithm Selection
3. Reporting on Accuracy and Efficiency

Prediction

1. Image Preprocessing
2. Image Classification

Imapge Processing Libraries

• http://scikit-image.org/
• https://opencv.org/

http://blog.mmast.net/python-image-processing-libraries-performance-opencv-scipy-scikit-image

Technical Design

How to Run?

Performance Data

Further Improvements

Applications