Object detection revamp

We are in the process of updating this repo with the latest research in the object detection community and would require collobrations on the following things

• re-implement from keras retinanet to mmdet, pytorch_lighting- retinanet
• write an introduction to mmdet.
• write an introduction to pytorch_lightning.
• revamp the object detection metrics section.
• train a few networks in various domains to benchmark: medical, retail, insurance, general
• future of object detection

while the ideas are vague at this point of time, we will make a detailed project plan in the coming days. Please raise issues and discuss in detail with the team before doing anything.