How use Pmosquito?#

Exigence#

To use this code, you need a CSV file in the following format:

• object: an integer identifier representing an object tracked over time.

• time: a decimal number representing the elapsed time in seconds.

• XSplined: the object's position along the X-axis (horizontal coordinate), expressed as a decimal number.

• YSplined: the object's position along the Y-axis (vertical coordinate), expressed as a decimal number.

• ZSplined: the object's position along the Z-axis (depth or height), expressed as a decimal number.

• VXSplined: the object's velocity along the X-axis, expressed as a decimal number.

• VYSplined: the object's velocity along the Y-axis, expressed as a decimal number.

• VZSplined: the object's velocity along the Z-axis, expressed as a decimal number.

Each row of the file therefore corresponds to the state of an object at a given moment, including its spatial position and velocity in all three dimensions.

Example (excerpt):

Project Structure#

The folder is organized as follows:

 PMosquito/
├── main.py                         # Main script
├── utils.py                        # Utility functions (clustering, calculations, visualizations)
├── requirements.txt                # Python dependencies
└── Tests
    ├── test_set_fictive.csv    
    ├── test_set_semi_fictionnal.csv
    ├── test_set_representing_a_tracking_bug.csv
    ├── test_set_extract_real_data.csv
    ├── result_test_set_fictive
        ├── test_set_fictive_reconstitue.csv 
        └── visualisation.png # swarm swarm visualization
    ├── result_test_set_semi_fictionnal
        ├── test_set_semi_fictionnal_reconstitue.csv 
        └── visualisation.png # swarm swarm visualization 
    └── result_test_set_extract_real_data
        ├── debug
            ├── connexions_spatiotemporelles.csv 
            ├── connexions_valides.csv 
            ├── matrice_spatiotemporelle.csv 
            └── PostProc_Filtered_2022_06_23_18_48_35_Splined_avec_features 
        ├── graphiques 
            ├── histogram_distance.png 
            ├── histogram_time.png 
            ├── mirrored_duration_histogram.png
            └── reconstitition_graphique.png 
        └── PostProc_Filtered_2022_06_23_18_48_35_Splined_reconstitue 

Tests files#

The test folder includes several different data files. - test_set_fictive.csv : This file contains 50 initial trajectories divided into 4 fragments. The times associated with the trajectories are between t= 0s and t=30s. Each trajectory is derived from functions allowing the creation of trajectories that can approximate those of a swarm of mosquitoes.

• test_set_semi_fictionnal.csv : This file contains 66 initial trajectories divided into 4 fragments. The times associated with the trajectories are between t= 0s and t=30s. Each trajectory is taken from real data but has been repositioned in time and selected because it lasted at least 30sec.

• test_set_representing_a_tracking_bug.csv : These are the same trajectories as the previous file, but the fragments are segmented 0.2 seconds to the 20th second for all trajectories. This file can be compared to a technical tracking bug.

• test_set_extract_real_data.csv : This dataset is an extract from a csv file that represents our real data from mosquito video tracking.

Output Files#

The program generates:

your_filename_reconstitute.csv # Data with updated trajectory identifiers (when a trajectory is considered a continuation of another)

With option debug :

connexions_spatiotemporelles.csv 
connexions_valides.csv # fragments of trajectories that come together
matrice_spatiotemporelle.csv # result
your_filename_avec_features # add features

With option graphiques :

histogram_distance.png # distance during the gap
histogram_time.png # gap time 
mirrored_duration_histogram.png # comparison of durations after reconstitution
reconstitition_graphique.png #  visual of the durations of the trajectories and their reconstructions

Available Parameters#

You can customize trajectory reconstruction using the following parameters:

Run Example#

Here's an example command to run the program:

pmosquito path_to_your_file.csv

Make sure you're always in the Pmosquito file: your terminal command line should always start with C:\Your_path_to\PMosquito\ > If this isn't the case, don't forget to

cd path_name

To add options, simply enter: "-- name of the option desired parameter"

C:\Your_path_to\PMosquito\ > python main.py path_to_your_file.csv --seuil_temps 0.4 --seuil_distance 0.2 --debug --time-min-reconstitute 10.0

In this example:

• Connected mosquitoes must be no more than 0.2m apart and within 0.4s of each other.

• The debug flag enables detailed logging and intermediate results.

• Trajectories shorter than 10 seconds are excluded from the final CSV output.