123456789101112131415161718 |
- _base_ = './point-rend_r50-caffe_fpn_ms-1x_coco.py'
- max_epochs = 36
- # learning policy
- param_scheduler = [
- dict(
- type='LinearLR', start_factor=0.001, by_epoch=False, begin=0, end=500),
- dict(
- type='MultiStepLR',
- begin=0,
- end=max_epochs,
- by_epoch=True,
- milestones=[28, 34],
- gamma=0.1)
- ]
- train_cfg = dict(max_epochs=max_epochs)
|