Skip to content

Latest commit

 

History

History
27 lines (19 loc) · 904 Bytes

README.md

File metadata and controls

27 lines (19 loc) · 904 Bytes

VisDrone2018

Task 1 : Object Detection in Images

detection

Environment

python3.5 tensorflow1.8.0 keras2.1.6

sudo pip3 install -r requirments.txt

Usage

python train_visdrone.py

TODO

  • Reading data and label using member function.(VisDroneDataset.load_VisDrone() and VisDroneDataset.load_anno().)
  • Writing configuration function.(VidDroneConfig, which is inheriting from class Config.)
  • Inspecting data.(load and display random 4 samples.)
  • Training network using the modified mask-rcnn algorithm based on VisDrone task.(train_visdrone.py)
  • Using Tensorboard to keep watch on training.
  • Test and submit results to leaderboard
  • Re-design the architecture of our network.

License

  • For academic and non-commercial use only.
  • For commercial use, please contact [email protected].