search

SwinTransformer / Swin-Transformer-Object-Detection

This is an official implementation for "Swin Transformer: Hierarchical Vision Transformer using Shifted Windows" on Object Detection and Instance Segmentation.
https://arxiv.org/abs/2103.14030
Apache License 2.0
1.81k stars 381 forks source link

convergence problem for Cascade Mask RCNN #10

Open GrassBro opened 3 years ago

GrassBro commented 3 years ago

Thanks for sharing your impressive work.

Recently, I train Swin-based Cascade Mask RCNN and Mask RCNN on my own dataset, including1.5k single-class images. The training process of Mask RCNN is relatively stable. However, the loss of Cascade Mask RCNN only converges to around 1.6 and doesn't decrease anymore. Its validation performance, meanwhile, is lower than the Mask RCNN counterpart.

They are trained in the same setting, could you give me some suggestions about fine-tuning the Cascade Mask RCNN?

HelyModi commented 3 years ago

Hi @TinyGrass , I am facing the same issue for Cascade Mask RCNN. Loss revolves around 1.5-1.7 and does not go lower than that.

kentaroy47 commented 3 years ago

Cascade trained well in my case. I used dist_train.sh for training and train.py did not work for cascade.

HelyModi commented 3 years ago

Cascade trained well in my case. I used dist_train.sh for training and train.py did not work for cascade.

Hey @kentaroy47, I used dist_train.sh as well, still no improvements.

kentaroy47 commented 3 years ago

Here is my config, hope it helps. I only changed the num_class in models and turned off fp16.

_base_ = [
    '_base_/datasets/coco_instance.py',
    '_base_/schedules/schedule_1x.py', 
    '_base_/default_runtime.py',
]
# model settings
model = dict(
    type='CascadeRCNN',
    pretrained=None,
    backbone=dict(
        type='SwinTransformer',
        embed_dim=96,
        depths=[2, 2, 18, 2],
        num_heads=[3, 6, 12, 24],
        window_size=7,
        mlp_ratio=4.,
        qkv_bias=True,
        qk_scale=None,
        drop_rate=0.,
        attn_drop_rate=0.,
        drop_path_rate=0.2,
        ape=False,
        patch_norm=True,
        out_indices=(0, 1, 2, 3),
        use_checkpoint=False),
    neck=dict(
        type='FPN',
        in_channels=[96, 192, 384, 768],
        out_channels=256,
        num_outs=5),
    rpn_head=dict(
        type='RPNHead',
        in_channels=256,
        feat_channels=256,
        anchor_generator=dict(
            type='AnchorGenerator',
            scales=[8],
            ratios=[0.5, 1.0, 2.0],
            strides=[4, 8, 16, 32, 64]),
        bbox_coder=dict(
            type='DeltaXYWHBBoxCoder',
            target_means=[.0, .0, .0, .0],
            target_stds=[1.0, 1.0, 1.0, 1.0]),
        loss_cls=dict(
            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
        loss_bbox=dict(type='SmoothL1Loss', beta=1.0 / 9.0, loss_weight=1.0)),
    roi_head=dict(
        type='CascadeRoIHead',
        num_stages=3,
        stage_loss_weights=[1, 0.5, 0.25],
        bbox_roi_extractor=dict(
            type='SingleRoIExtractor',
            roi_layer=dict(type='RoIAlign', output_size=7, sampling_ratio=0),
            out_channels=256,
            featmap_strides=[4, 8, 16, 32]),
        bbox_head=[
            dict(
                type='ConvFCBBoxHead',
                num_shared_convs=4,
                num_shared_fcs=1,
                in_channels=256,
                conv_out_channels=256,
                fc_out_channels=1024,
                roi_feat_size=7,
                num_classes=4,
                bbox_coder=dict(
                    type='DeltaXYWHBBoxCoder',
                    target_means=[0., 0., 0., 0.],
                    target_stds=[0.1, 0.1, 0.2, 0.2]),
                reg_class_agnostic=False,
                reg_decoded_bbox=True,
                norm_cfg=dict(type='SyncBN', requires_grad=True),
                loss_cls=dict(
                    type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0),
                loss_bbox=dict(type='GIoULoss', loss_weight=10.0)),
            dict(
                type='ConvFCBBoxHead',
                num_shared_convs=4,
                num_shared_fcs=1,
                in_channels=256,
                conv_out_channels=256,
                fc_out_channels=1024,
                roi_feat_size=7,
                num_classes=4,
                bbox_coder=dict(
                    type='DeltaXYWHBBoxCoder',
                    target_means=[0., 0., 0., 0.],
                    target_stds=[0.05, 0.05, 0.1, 0.1]),
                reg_class_agnostic=False,
                reg_decoded_bbox=True,
                norm_cfg=dict(type='SyncBN', requires_grad=True),
                loss_cls=dict(
                    type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0),
                loss_bbox=dict(type='GIoULoss', loss_weight=10.0)),
            dict(
                type='ConvFCBBoxHead',
                num_shared_convs=4,
                num_shared_fcs=1,
                in_channels=256,
                conv_out_channels=256,
                fc_out_channels=1024,
                roi_feat_size=7,
                num_classes=4,
                bbox_coder=dict(
                    type='DeltaXYWHBBoxCoder',
                    target_means=[0., 0., 0., 0.],
                    target_stds=[0.033, 0.033, 0.067, 0.067]),
                reg_class_agnostic=False,
                reg_decoded_bbox=True,
                norm_cfg=dict(type='SyncBN', requires_grad=True),
                loss_cls=dict(
                    type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0),
                loss_bbox=dict(type='GIoULoss', loss_weight=10.0))

        ],
        mask_roi_extractor=dict(
            type='SingleRoIExtractor',
            roi_layer=dict(type='RoIAlign', output_size=14, sampling_ratio=0),
            out_channels=256,
            featmap_strides=[4, 8, 16, 32]),
        mask_head=dict(
            type='FCNMaskHead',
            num_convs=4,
            in_channels=256,
            conv_out_channels=256,
            num_classes=4,
            loss_mask=dict(
                type='CrossEntropyLoss', use_mask=True, loss_weight=1.0))),
    train_cfg=dict(
        rpn=dict(
            assigner=dict(
                type='MaxIoUAssigner',
                pos_iou_thr=0.7,
                neg_iou_thr=0.3,
                min_pos_iou=0.3,
                match_low_quality=True,
                ignore_iof_thr=-1),
            sampler=dict(
                type='RandomSampler',
                num=256,
                pos_fraction=0.5,
                neg_pos_ub=-1,
                add_gt_as_proposals=False),
            allowed_border=-1,
            pos_weight=-1,
            debug=False),
        rpn_proposal=dict(
            nms_pre=2000,
            max_per_img=1000,
            nms=dict(type='nms', iou_threshold=0.7),
            min_bbox_size=0),
        rcnn=dict(
            assigner=dict(
                type='MaxIoUAssigner',
                pos_iou_thr=0.5,
                neg_iou_thr=0.5,
                min_pos_iou=0.5,
                match_low_quality=True,
                ignore_iof_thr=-1),
            sampler=dict(
                type='RandomSampler',
                num=512,
                pos_fraction=0.25,
                neg_pos_ub=-1,
                add_gt_as_proposals=True),
            mask_size=28,
            pos_weight=-1,
            debug=False)),
    test_cfg=dict(
        rpn=dict(
            nms_pre=1000,
            max_per_img=1000,
            nms=dict(type='nms', iou_threshold=0.7),
            min_bbox_size=0),
        rcnn=dict(
            score_thr=0.001,
            nms=dict(type='nms', iou_threshold=0.5),
            max_per_img=100,
            mask_thr_binary=0.5)))

img_norm_cfg = dict(
    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)

# augmentation strategy originates from DETR / Sparse RCNN
train_pipeline = [
    dict(type='LoadImageFromFile'),
    dict(type='LoadAnnotations', with_bbox=True, with_mask=True),
    dict(type='RandomFlip', flip_ratio=0.5),
    dict(type='AutoAugment',
         policies=[
             [
                 dict(type='Resize',
                      img_scale=[(480, 1333), (512, 1333), (544, 1333), (576, 1333),
                                 (608, 1333), (640, 1333), (672, 1333), (704, 1333),
                                 (736, 1333), (768, 1333), (800, 1333)],
                      multiscale_mode='value',
                      keep_ratio=True)
             ],
             [
                 dict(type='Resize',
                      img_scale=[(400, 1333), (500, 1333), (600, 1333)],
                      multiscale_mode='value',
                      keep_ratio=True),
                 dict(type='RandomCrop',
                      crop_type='absolute_range',
                      crop_size=(384, 600),
                      allow_negative_crop=True),
                 dict(type='Resize',
                      img_scale=[(480, 1333), (512, 1333), (544, 1333),
                                 (576, 1333), (608, 1333), (640, 1333),
                                 (672, 1333), (704, 1333), (736, 1333),
                                 (768, 1333), (800, 1333)],
                      multiscale_mode='value',
                      override=True,
                      keep_ratio=True)
             ]
         ]),
    dict(type='Normalize', **img_norm_cfg),
    dict(type='Pad', size_divisor=32),
    dict(type='DefaultFormatBundle'),
    dict(type='Collect', keys=['img', 'gt_bboxes', 'gt_labels', 'gt_masks']),
]
data = dict(train=dict(pipeline=train_pipeline))

optimizer = dict(_delete_=True, type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05,
                 paramwise_cfg=dict(custom_keys={'absolute_pos_embed': dict(decay_mult=0.),
                                                 'relative_position_bias_table': dict(decay_mult=0.),
                                                 'norm': dict(decay_mult=0.)}))