Generation of multiple annotations at once

Project-AgML / AgML

AgML is a centralized framework for agricultural machine learning. AgML provides access to public agricultural datasets for common agricultural deep learning tasks, with standard benchmarks and pretrained models, as well the ability to generate synthetic data and annotations.

Apache License 2.0

167 stars 28 forks source link

Generation of multiple annotations at once #43

Open ctyeong opened 1 year ago

ctyeong commented 1 year ago

For synthesis, I'd like to suggest a feature to generate multiple types of annotation–bounding box, instance/class segmentation, etc.–per image at once. If I've understood correctly, the current version of AgML can only generate one type, and I need to synthesize again, if I need another type. It costs time, and the synthesized environment can change.

amogh7joshi commented 1 year ago

Hi Taeyong, this isn't due to AgML, this is due to the way the underlying Helios generation works. However, I can take a look into this and see if there's a way to generate multiple types of annotations at once.

ctyeong commented 1 year ago

Thanks @amogh7joshi I've just found that we can make a simple tweak on converter.py to generate instance segmentation while using object detection annotation. An examples is below. I'll do pull request if this'd help others as well.

masonearles commented 1 year ago

Thanks @ctyeong! Yes please make a PR.