MVP: Implement ML model trainer for threat detection based on YOLO

[Chatura]

Tasks

1. Decide the YOLO version
2. Finalize the data annotation format.
3. Create a training dataset
4. Implement the model trainer

[Chatura]

1. YoloV8 was selected as the final model.
2. Ultralytics YOLO format was finalized as the data annotation format.
3. The following 2 datasets are used for initial model
   a. Gun Detection V1 dataset (~300 datapoints)
   b. Weapon 2 dataset (~ 3900 datapoints)