I've no idea how to work with this

[AzethMeron]

Issue Description
Hello, I've no experience yet with object detection or segmentation when it comes to designing or training such NN. I've tried hard, but I've failed to use this repo. Huge chunk of docs and sample notebooks are blank/todos.

Can anyone please make a recipe how to train yolo on coco, create yolo object in python and load pretrained weights into this object?

I know documentation isn't a priority in ongoimg open source project but please, anyone.

[Erotemic]

I'm working on a fork, which contains an end-to-end example of generating a demo dataset, training a model, and evaluating with it.

#150

With the script here:
https://github.com/MultimediaTechLab/YOLO/blob/8d7a738fb16c6713eb54458583b7c9c7c41c0ee5/train_kwcoco_demo.sh

This relies on kwcoco.

I'm also still working out exactly how to use this library more effectively, but I'm hoping some of my improvements can eventually be mainlined.