Trajectories-Prediction-Kalman

Study and analysis of trajectories Prediction with Kalman Filter. The file 'report.pdf' describes the work.

Installation

The scripts are written in Python 3.6.

This project requires the following Python packages installed:

• numpy
• matplotlib

Example execution

This command starts the trajectories prediction analysis using kalman filter with uniformly accelerated motion and save the qualitative results:

$ python main.py -s -a 

The details of analysis and qualitative results are saved in a folder.

For analysis described in chapter 5.4 in report.pdf, use the command:

$ python analysis_homography

Before, you must extract "dataset_trajectories_frame.zip" containing "dataset_tracjetories_small.json"

Note: the analysis ( chapter 5.4 ) is made for a single sequence of dataset KITTI ( 0018 ).

Command line arguments

    -h, --help                     show this help message and exit.
    -s, --save                     save the qualitative results.
    -p0 P0                         P0 diagonal value, the initial Process Covariance Matrix. (default: 0.03)
    -q Q                           Q diagonal value, the Process Noise Covariance Matrix. (default: 0.03)
    -r0 R0                         R0 diagonal value, the initial Measurements Noise Covariance Matrix. (default: 0.03)
    --past_len                     past length (default: 20)
    --future_len                   future length (default: 40)
    -a, --acceleration             use acceleration (default: False)

Note: This project has been developed for the course "Image and Video Analysis" ( Università degli studi di Firenze ). It has been resumed the work made by Simone Magistri and Ivan Prosperi. I thank them for excellent work.

Authors

• Francesco Marchetti