loaiabdalslam/ml_traffic_detection_tracking

recognizing vehicles and traffic signs and tracking them

recognizing vehicles and traffic signs and tracking them

Table of contents

• How it works
• Prerequisites
• Getting started
  • image_process.py
  • seq_image_process.py
  • vid_process.py

How it works

This code simply feeds yolo trained model into cv2 deep neural network

  net = cv2.dnn.readNetFromDarknet(configPath, weightsPath)

and track the detected objects using sort

Prerequisites

• download yolov3.weights and put it in yolo-coco folder
• make sure you have these libs installed:
  • numpy
  • imutils
  • opencv

Getting started

There is actually three different script, each doing a distinguished task.

image_process.py

This script is used to process images that are not necessarily taken in a sequence (images taken in different times). It only detects objects but no tracking is done.

Simply, put your images in input/image/. Run

  python3 image_process.py

It will output:

• processed images (images with boxes and text) in output.
• in det_results/, text file for each image containing detected classes in this format:

classID confidence left top right bottom

seq_image_process.py

This script is used to process images taken in a sequence (like video frames). It detects objects and track them.

Put your images in input/image/. Run

  python3 seq_image_process.py

It will output:

• processed frames (images with boxes and text) in output.
• one CSV file with a line format like follows:

frame_index DetectionIndex_ClassName right top left bottom

vid_process.py

This script is used to process videos. It detects objects and track them.

Rename your video to input.mp4 and move it to input/video. Run

  python3 vid_process.py

It will output:

• processed frames (images with boxes and text) in output.
• processed video in AVI format named output.avi.
• one CSV file with a line format like follows:

frame_index DetectionIndex_ClassName right top left bottom

Citation

YOLO :

@article{redmon2016yolo9000,
  title={YOLO9000: Better, Faster, Stronger},
  author={Redmon, Joseph and Farhadi, Ali},
  journal={arXiv preprint arXiv:1612.08242},
  year={2016}
}

SORT :

@inproceedings{Bewley2016_sort,
  author={Bewley, Alex and Ge, Zongyuan and Ott, Lionel and Ramos, Fabio and Upcroft, Ben},
  booktitle={2016 IEEE International Conference on Image Processing (ICIP)},
  title={Simple online and realtime tracking},
  year={2016},
  pages={3464-3468},
  keywords={Benchmark testing;Complexity theory;Detectors;Kalman filters;Target tracking;Visualization;Computer Vision;Data Association;Detection;Multiple Object Tracking},
  doi={10.1109/ICIP.2016.7533003}
}