README.md

C6 Team 5 - Week 2

Data

Add the data to the a folder named data in the root of the project. The data should be organized as follows:

data
│
└───AICity_data
|   |
|   └───train
|   |   └───S03
|   |   |   └───c010
|   |   |   |   └───vdo.avi
└───ai_challenge_s03_c010-full_annotation.xml

Preparation

To install the required packages to run the program, execute the following command:

pip install -r requirements3_10.txt

• We reccomend to use Python 3.10 to run the program.
• We recomend to use a virtual environment to avoid conflicts with other projects.

Also run the following command to create the frame_dict.json file, containing the labels in a more convenient format:

cd scripts
python proc_xml_to_json.py

Execution

Training and tuning

To train the model and tune the hyperparameters, execute one of the following commands:

cd scripts
python yolo_train.py
python yolo_tune.py

To specify the dataset configuration that YOLO usually needs, refer to the dataset.yaml file.

Object and tracking

To compute the pipeline for the object tracking just execute the following command to run the program choosing the desired tracking method:

python main.py [--tag TAG][--tracking-method {overlap,kalman_sort}] [--show-tracking] [--save-for-track-eval][--frames-percentage PERCENTAGE]

With the following options:
--tag TAG Tag for the output folder
--tracking-method Choose the tracking method --show-tracking If specified, show the tracking results
--save-for-track-eval If specified, save the tracking results in a format that can be used for the tracking evaluation
--frames-percentage Percentage of frames to skip while showing the video of the tracking results

Also, indicate paths for model to use in file of global variables.