This assignment project aims to recognize the type of moving objects in video footage. The realization is composed of two parts:
-
Video Analysis and ROI Extraction
The video footage is analyzed using the OpenCV API to extract the regions of interest (ROI) of moving objects. -
Model Training and Prediction
- Transfer Learning: A transfer learning model is retrained and fine-tuned using an open-source vehicle image dataset.
- Prediction: The fine-tuned transfer learning model is used to predict the extracted ROIs of moving objects.
- Ensemble Learning: Majority voting is applied to enhance the robustness of the prediction model.