TEA stands for Traffic extraction and analyzation. This is a tool which provides with the future trajectory of vehicles, shows their trajectories along with characteristics such as position, velocity and acceleration.The color detection is done using KNN. The object detection is based on YOLOv3 and the trajectory prediction is based on Kalman Filter.
- OpenCV
- Yolo V3
- Python
- Kalman Filter
The tool is extreamly simple to use. The user has to just update the video or image.
- The user can choose from video as well as image input. The commands are given as
python3 main.py --video path python3 main.py --image path - The output is stored as filename.output.extension and the csv dump is found as data.csv
- The csv file has the data as the x-coordinate, y-coordinate,speed, acceleration and the object type of the vehicle. Each of the vehicle as a unique vehicle id and the data for vehicles is seperate by new line in the csv.
A sample shot of a csv file
The extraction and analyzation in action. The series of blue points is the path traced by the object. The green lines is the trajectory estimated using kalmaan filter. The box has color detected using KNN and the vehicle type using Yolo V3.
