centernet-update_r50_fpn_8xb8-amp-lsj-200e_coco.py

_base_ = '../common/lsj-100e_coco-detection.py'

image_size = (1024, 1024)
batch_augments = [dict(type='BatchFixedSizePad', size=image_size)]
max_epochs = 25

model = dict(
    type='CenterNet',
    data_preprocessor=dict(
        type='HighDetDataPreprocessor',
        # mean=[0.0] * 10,
        mean=None,
        # std=[1.0] * 10,
        std=None,
        bgr_to_rgb=True,
        pad_size_divisor=32,
        batch_augments=batch_augments),
    stem=dict(
        type='RGBHSIStem',
        in_channels=10,
        out_channels=64,
    ),
    backbone=dict(
        type='ResNet',
        in_channels = 10,
        depth=50,
        num_stages=4,
        out_indices=(0, 1, 2, 3),
        frozen_stages=1,
        norm_cfg=dict(type='BN', requires_grad=True),
        norm_eval=True,
        style='pytorch',
        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
    neck=dict(
        type='FPN',
        in_channels=[256, 512, 1024, 2048],
        out_channels=256,
        start_level=1,
        add_extra_convs='on_output',
        num_outs=5,
        init_cfg=dict(type='Caffe2Xavier', layer='Conv2d'),
        relu_before_extra_convs=True),
    bbox_head=dict(
        type='CenterNetUpdateHead',
        num_classes=5,
        in_channels=256,
        stacked_convs=4,
        feat_channels=256,
        strides=[8, 16, 32, 64, 128],
        loss_cls=dict(
            type='GaussianFocalLoss',
            pos_weight=0.25,
            neg_weight=0.75,
            loss_weight=1.0),
        loss_bbox=dict(type='GIoULoss', loss_weight=2.0),
    ),
    
    train_cfg = dict(
        type='EpochBasedTrainLoop', max_epochs=max_epochs, val_interval=5),
    test_cfg=dict(
        nms_pre=1000,
        min_bbox_size=0,
        score_thr=0.05,
        nms=dict(type='nms', iou_threshold=0.6),
        max_per_img=100))


# Enable automatic-mixed-precision training with AmpOptimWrapper.
optim_wrapper = dict(
    type='AmpOptimWrapper',
    optimizer=dict(
        type='SGD', lr=1e-4, momentum=0.9, weight_decay=0.00004),
    paramwise_cfg=dict(norm_decay_mult=0.))

param_scheduler = [
    dict(
        type='LinearLR',
        start_factor=0.00025,
        by_epoch=False,
        begin=0,
        end=4000),
    dict(
        type='MultiStepLR',
        begin=0,
        end=25,
        by_epoch=True,
        milestones=[22, 24],
        gamma=0.1)
]

# NOTE: `auto_scale_lr` is for automatically scaling LR,
# USER SHOULD NOT CHANGE ITS VALUES.
# base_batch_size = (8 GPUs) x (8 samples per GPU)
auto_scale_lr = dict(base_batch_size=8)
train_dataloader = dict(batch_size=8, num_workers=8)