Build-in ML models' performance

[ZhiruiFeng]

The project uses existing Deep Learning models for object detection and scene detection. Fine-tuning ML models are not within the scope of Data Engineering, thus the first version of this project doesn't consider the performance of these modules. For example, current object detection could not work on dark night frames.

However, this project's architecture decouples every processing modules and treats ML modules as API, which makes it very flexible to adjust and change to new models.

We welcome users and contributors to put customized detection modules in src.processing.objdetector.py and src.processing.scenedetector.py.