Pothole Severity Classification via Computer Vision

The Project was build for the SMARTATHON Challenge 2023 theme 2 (Pothole Severity Classification via Computer Vision).

Instructions to run the code

Run this notebook.

        (or)           

Run it locally with the following instructions:

• Clone the repository.

git clone https://github.com/shivankar-p/pothole.git

• Go into the repository

cd pothole/src

• Linux - Install metashape standalone module wheel File.(For other os find wheel file here

wget https://s3-eu-west-1.amazonaws.com/download.agisoft.com/Metashape-2.0.0-cp35.cp36.cp37.cp38-abi3-linux_x86_64.whl
python3 -m pip install Metashape-2.0.0-cp35.cp36.cp37.cp38-abi3-linux_x86_64.whl

• Install other required packages

pip install -r requirements.txt

• Inside the models folder(in src) download the model file from this link
• Navigate back to src directory
• For Inference on image run:

python3 mask.py --image <image-path>

• For Inference on video run:

python3 mask.py --video <video-path>

Technologies used

• python
• Agiosoft Metashape
• openCV
• tensorflow
• keras
• matplotlib
• numpy

Demo video link

Output video link

Output Screenshots

Team

• Shivankar Pilligundla
• Hardik Khandelwal
• Karanjit Saha
• Netradeepak Chinchwadkar
• Abhinav Mahajan
• Srinivas Manda