We provide Faster R-CNN and RetinaNet results on Tiny-DOTA datasets, and a script for drawing number of clicks (NoC) graph.
python draw_noc_graph.py
Note that you can obtain new models' NoC curve.
- Add new model results as XXX.json files in the Tiny-DOTA-FasterRCNN or Tiny-DOTA-RetinaNet folder.
- Change code in the
draw_noc_graph.py
as follow:
files = ['Ours.json', 'Late_Fusion.json', 'Early_Fusion.json', 'Passthrough.json', 'Faster_R-CNN.json']
Each json file represents our model and baseline results which included in average of mAP and standard deviation calculated by five times inference results.
Each json file represents our model and baseline results which included in average of mAP and standard deviation calculated by five times inference results.