This repository contains a Python script for tracking objects in a video using a particle filter-based method known as the Condensation algorithm. The tracker estimates the position of a bounding box over time by propagating particles, observing weights, and resampling based on observed data. It also allows for the visualization of the tracking process.
- Object Tracking: Track objects across frames using a particle filter.
- Histogram-based Observation: Utilizes color histograms to observe and update the particle weights.
- Interactive Bounding Box Selection: Allows users to select the initial bounding box for tracking.
- Visualization: Provides a visual representation of the tracking process.
- Install all the requirements from requirements.txt (I am using Python 3.8.18)
- Run
python code/condensation_tracker.py
- Choose a starting bounding box for the object to be tracked and then press 'q'
- Press 'q' to exit the window
Contributions are welcome! Feel free to contribute by implementing new features, improvements, or bug fixes . Please open an issue or pull request on GitHub.