OpenCV 3 & Keras implementation of vehicles tracking with video data.
Algorithm
- Object Detection: MOG2
- Object Tracking: KCF
- Object Classification: CNN
Requirement
- Python 3.6
- OpenCV 3.2 + contrib
- Tensorflow-gpu 1.0
- Keras 1.2
We train our CNN model with MIT's vehicle and pedestrian data, click here to download the original data and the processed data.
Video data is saved in the video folder.
The CNN model we use is as follows：
Run the following command to execute the program.
python track.py --file "car.flv"
