UAV-traffic-analysis

This project aims to count every vehicle (bus, car, truck) detected in the input video using YOLOv5 object-detection algorithm. It also calculates speed, acceleration of every vehicle

DEMO

Steps to run

1. Clone this repository

   git clone https://github.com/Garvit-32/UAV.git
   

2. Install all the dependencies

   pip install -r requirements.txt   
   

3. Collect 6-8 coordinates points (latitude, longitude) from google maps and (x, y) from the image and put it in Code/localization/localization_from_points.py from there you get the pixel delta (delta x, delta y) which is used to calculate distance, speed and acceleration of vehicle later.

   

   In the given script, 8 points of this image are given, likewise you have to collect and put.

4. Exceute main script to start tracking

   python Code/main.py -p <path to video> -o <latitude and longitude coordinate of the center of the frame > -pd < pixel delta calculated from localization script>  
   

   Example

   python Code/main.py -p Input/DJI_0004_gt.mp4 -o "37.47646052, 126.89894" -pd "-2.00e-06, -2.00e-06"
   

5. Evaluate Tracking (GroudTruth are required)

   python Code/tracking_eval/tracking_eval.py -p <Path to folder containing groundtruth> -v <video name> 
   

   Example

   python Code/tracking_eval/tracking_eval.py -p Input/yolo_id -v DJI_0004_gt
   

   Tracking result for the given video are:

   MOTA	0.7994
   MOTP	0.7660
   Ground Truths	3884
   False Positives	757
   Misses	10
   Mismatches	12
   Correspondences	3874

References

1. https://github.com/ultralytics/yolov5
2. https://github.com/Videmo/pymot