search

Cartucho / OpenLabeling

Label images and video for Computer Vision applications
Apache License 2.0
926 stars 266 forks source link

Export a package of input & output for yolo_darknet training #44

Closed jamietownsend closed 5 years ago

jamietownsend commented 5 years ago

Merge conflicts resolved

Cartucho commented 5 years ago

Hello! Thanks, so the export is a way to copy the files automatically to the yolo_darknet folder and set things up for training?

jamietownsend commented 5 years ago

Exactly. It collects the images and annotations together with the obj.names file as required to use https://github.com/AlexeyAB/darknet

jamietownsend commented 5 years ago

Hi Cartucho, any advance on this one?

Cartucho commented 5 years ago

Hello, I like the idea and I think we should merge it, however, we need to add some flexibility for other future types of exports.

I think we should use two parameters: python main.py --export [export_type] [export_dir_path]

Because one does not export to darknet necessarily, they could be using darkflow for example and in this case the files to be copied are the .xmlfiles instead of the .txt

Additionally, elif pressed_key == ord('x'): should only be triggered if the person indeed activated the --export flag.