This project demonstrates real-time object tracking using OpenCV's built-in tracking algorithms such as KCF, CSRT, and MIL. You can track objects from a webcam or video file and view FPS, success status, and execution time per frame.
- 📹 Supports external/webcam and video file inputs
- 🎯 Real-time object tracking with multiple OpenCV algorithms
- 🕒 Displays FPS and total execution time
- 📈 Prints performance summary at the end
- ✅ ROI (Region of Interest) selector for target object
csrt- Discriminative Correlation Filter with Channel and Spatial Reliabilitykcf- Kernelized Correlation Filtersmil- Multiple Instance Learning
Other trackers (
tld,mosse,boosting,medianflow) may not work with OpenCV ≥ 4.11 due to API changes.
git clone https://github.com/praveensunkara19/ObjectTracker.git
cd ObjectTracker
2. Set up virtual environment (optional but recommended)
```bash
python -m venv myenv
myenv\Scripts\activate # On Windows
3. Install dependencies
```bash
pip install -r requirements.txt
Note: Use opencv-contrib-python for tracker support.
Dependencies
Python 3.8+
opencv-contrib-python==4.5.5.64
imutils
numpy==1.23.5for csrt:
python ObjectTracker_csrt.pyfor - csrt | kcf | mil Using Webcam:
python ObjectTracker.py --tracker kcfUsing Video:
python ObjectTracker.py --video path/to/video.mp4 --tracker csrt #test_car.mp4During Execution: Press S to select an object to track Press Q to quit the application
📊 Tracking Summary Output At the end of tracking, you will get:
-- [SUMMARY] "Tracker Total frames tracked: 100 Total tracking execution time: 4.2156 seconds Average tracker.update() time: 0.042156 seconds Total real time spent (including display, etc.): 5.1204 seconds Sample Output Frame
🙋 Author Praveen Sunkara GitHub: @praveensunkara19 Email: praveesunkara19@gmail.com