ddrnet: The size of tensor a (182) must match the size of tensor b (181) at non-singleton dimension 3

[yazansayed]

I looked briefly at the issues and it seems the problem occurs with other methods as well, I'm using a crop size of (1024,1024) and scale = (2048,1024), all divisible by 32.
the error occurs at evaluation and inference:

[/content/mmsegmentation/mmseg/models/backbones/ddrnet.py](https://localhost:8080/#) in forward(self, x)
    191         comp_c = self.compression_1(self.relu(x_c))
    192         x_c += self.down_1(self.relu(x_s))
--> 193         x_s += resize(
    194             comp_c,
    195             size=out_size,

RuntimeError: The size of tensor a (182) must match the size of tensor b (181) at non-singleton dimension 3

here's the config:

crop_size = (832, 832)
scale = (2048, 832)
test_scale = (1024, 1024)
data_root = 'dataxseg'
dataset_type = 'TextDet1Dataset'
save_interval = 500
max_iters = 6000
val_interval= 500
load_from = None
batch_size=16
startlr = 0.0005
min_lr=1e-06
num_workers=2


class_weight = [
    0.8, 1.1
]
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/ddrnet/pretrain/ddrnet23s-in1kpre_3rdparty-1ccac5b1.pth'  # noqa
data_preprocessor = dict(
    type='SegDataPreProcessor',
    size=crop_size,
    mean=[123.675, 116.28, 103.53],
    std=[58.395, 57.12, 57.375],
    bgr_to_rgb=True,
    pad_val=0,
    seg_pad_val=255)
norm_cfg = dict(type='SyncBN', requires_grad=True)
model = dict(
    type='EncoderDecoder',
    data_preprocessor=data_preprocessor,
    backbone=dict(
        type='DDRNet',
        in_channels=3,
        channels=32,
        ppm_channels=128,
        norm_cfg=norm_cfg,
        align_corners=False,
        init_cfg=dict(type='Pretrained', checkpoint=checkpoint)),
    decode_head=dict(
        type='DDRHead',
        in_channels=32 * 4,
        channels=64,
        dropout_ratio=0.,
        num_classes=2,
        align_corners=False,
        norm_cfg=norm_cfg,
        loss_decode=[
            dict(
                type='OhemCrossEntropy',
                thres=0.9,
                min_kept=131072,
                class_weight=class_weight,
                loss_weight=1.0),
            dict(
                type='OhemCrossEntropy',
                thres=0.9,
                min_kept=131072,
                class_weight=class_weight,
                loss_weight=0.4),
        ]),

    # model training and testing settings
    train_cfg=dict(),
    test_cfg=dict(mode='whole'))

train_dataloader = dict(batch_size=batch_size, num_workers=num_workers)

# optimizer
# optimizer = dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0005)
optimizer = dict(lr=startlr, type='AdamW', weight_decay=0.0005)

optim_wrapper = dict(type='OptimWrapper', optimizer=optimizer, clip_grad=None)
# learning policy
param_scheduler = [
    dict(
        type='PolyLR',
        eta_min=min_lr,
        power=0.9,
        begin=0,
        end=max_iters,
        by_epoch=False)
]

# training schedule for 120k
train_cfg = dict(
    type='IterBasedTrainLoop', max_iters=max_iters, val_interval=val_interval)
val_cfg = dict(type='ValLoop')
test_cfg = dict(type='TestLoop')
default_hooks = dict(
    timer=dict(type='IterTimerHook'),
    logger=dict(type='LoggerHook', interval=50, log_metric_by_epoch=False),
    param_scheduler=dict(type='ParamSchedulerHook'),
    checkpoint=dict(
        type='CheckpointHook', by_epoch=False, interval=save_interval),
    sampler_seed=dict(type='DistSamplerSeedHook'),
    visualization=dict(type='SegVisualizationHook'))

randomness = dict(seed=304)

test_cfg = dict(type='TestLoop')
test_dataloader = dict(
    batch_size=1,
    dataset=dict(
        data_prefix=dict(
            img_path='images/validation',
            seg_map_path='annotations/validation'),
        data_root=data_root,
        pipeline=[
            dict(type='LoadImageFromFile'),
            dict(keep_ratio=True, scale=scale, type='Resize'),
            dict(reduce_zero_label=False, type='LoadAnnotations'),
            dict(type='PackSegInputs'),
        ],
        type=dataset_type),
    num_workers=2,
    persistent_workers=True,
    sampler=dict(shuffle=False, type='DefaultSampler'))
test_evaluator = dict(
    iou_metrics=[
        'mIoU',
    ], type='IoUMetric')
test_pipeline = [
    dict(type='LoadImageFromFile'),
    dict(keep_ratio=True, scale=test_scale, type='Resize'),
    dict(reduce_zero_label=False, type='LoadAnnotations'),
    dict(type='PackSegInputs'),
]

train_cfg = dict(
    max_iters=max_iters, type='IterBasedTrainLoop', val_interval=val_interval)
train_dataloader = dict(
    batch_size=batch_size,
    dataset=dict(
        data_prefix=dict(
            img_path='images/training', seg_map_path='annotations/training'),
        data_root=data_root,
        pipeline=[
            dict(type='LoadImageFromFile'),
            dict(reduce_zero_label=False, type='LoadAnnotations'),
            dict(
                keep_ratio=True,
                ratio_range=(
                    0.5,
                    2.0,
                ),
                scale=scale,
                type='RandomResize'),
            dict(
                cat_max_ratio=0.75, crop_size=crop_size, type='RandomCrop'),
            dict(prob=0.5, type='RandomFlip'),
            dict(type='PhotoMetricDistortion'),
            dict(type='PackSegInputs'),
        ],
        type=dataset_type),
    num_workers=2,
    persistent_workers=True,
    sampler=dict(shuffle=True, type='InfiniteSampler'))
train_pipeline = [
    dict(type='LoadImageFromFile'),
    dict(reduce_zero_label=False, type='LoadAnnotations'),
    dict(
        keep_ratio=True,
        ratio_range=(
            0.5,
            2.0,
        ),
        scale=scale,
        type='RandomResize'),
    dict(cat_max_ratio=0.75, crop_size=crop_size, type='RandomCrop'),
    dict(prob=0.5, type='RandomFlip'),
    dict(type='PhotoMetricDistortion'),
    dict(type='PackSegInputs'),
]
tta_model = dict(type='SegTTAModel')
tta_pipeline = [
    dict(backend_args=None, type='LoadImageFromFile'),
    dict(
        transforms=[
            [
                dict(keep_ratio=True, scale_factor=0.5, type='Resize'),
                dict(keep_ratio=True, scale_factor=0.75, type='Resize'),
                dict(keep_ratio=True, scale_factor=1.0, type='Resize'),
                dict(keep_ratio=True, scale_factor=1.25, type='Resize'),
                dict(keep_ratio=True, scale_factor=1.5, type='Resize'),
                dict(keep_ratio=True, scale_factor=1.75, type='Resize'),
            ],
            [
                dict(direction='horizontal', prob=0.0, type='RandomFlip'),
                dict(direction='horizontal', prob=1.0, type='RandomFlip'),
            ],
            [
                dict(type='LoadAnnotations'),
            ],
            [
                dict(type='PackSegInputs'),
            ],
        ],
        type='TestTimeAug'),
]
val_cfg = dict(type='ValLoop')
val_dataloader = dict(
    batch_size=1,
    dataset=dict(
        data_prefix=dict(
            img_path='images/validation',
            seg_map_path='annotations/validation'),
        data_root=data_root,
        pipeline=[
            dict(type='LoadImageFromFile'),
            dict(keep_ratio=True, scale=scale, type='Resize'),
            dict(reduce_zero_label=False, type='LoadAnnotations'),
            dict(type='PackSegInputs'),
        ],
        type=dataset_type),
    num_workers=2,
    persistent_workers=True,
    sampler=dict(shuffle=False, type='DefaultSampler'))
val_evaluator = dict(
    iou_metrics=[
        'mIoU',
    ], type='IoUMetric')


img_ratios = [
    0.5,
    0.75,
    1.0,
    1.25,
    1.5,
    1.75,
]


default_scope = 'mmseg'
env_cfg = dict(
    cudnn_benchmark=True,
    mp_cfg=dict(mp_start_method='fork', opencv_num_threads=0),
    dist_cfg=dict(backend='nccl'),
)
vis_backends = [dict(type='LocalVisBackend')]
visualizer = dict(
    type='SegLocalVisualizer', vis_backends=vis_backends, name='visualizer')
log_processor = dict(by_epoch=False)
log_level = 'INFO'
resume = False

tta_model = dict(type='SegTTAModel')

[HePengguang]

I met the same issue, did u sovle this?

[Misaka19798]

I met the same bug, seems the Resize in tta caused this dimension mismatch. I simplely remove the tta since I don't need it, and the the eval runs well without tta.

[hoyeoung]

I met the same bug, seems the Resize in tta caused this dimension mismatch. I simplely remove the tta since I don't need it, and the the eval runs well without tta.

I met the same issue and I can't solve it. I'm sorry but can you show your config file??

[Sunhil]

我也遇到了这个问题，我定位到的原因是‘mmseg/models/backbones/ddrnet.py’ 中的stage3中的out_size和x_s的size不同。解决方案是：修改‘mmseg/models/backbones/ddrnet.py’ 文件。

...
import math #添加
...
# out_size = (x.shape[-2] // 8, x.shape[-1] // 8)  # 注释或删掉183行
out_size = (math.ceil(x.shape[-2] / 8), math.ceil(x.shape[-1] / 8)) # 改为向上取整