swin-l 加载权重后mAP很低 #13
Description
请问为什么我用swin-l加载权重后训练自己的数据集，mAP会很低，上不去，而在r50主干还有mm自带的swin-tiny训练就正常？
下面是日志：
2025-06-17 21:56:32,970 - mmrotate - INFO - Environment info:
sys.platform: linux
Python: 3.8.10 (default, Jun 4 2021, 15:09:15) [GCC 7.5.0]
CUDA available: True
GPU 0: NVIDIA GeForce RTX 4090
CUDA_HOME: /usr/local/cuda
NVCC: Cuda compilation tools, release 11.8, V11.8.89
GCC: gcc (Ubuntu 9.4.0-1ubuntu1~20.04.1) 9.4.0
PyTorch: 2.0.0+cu118
PyTorch compiling details: PyTorch built with:
- GCC 9.3
- C++ Version: 201703
- Intel(R) oneAPI Math Kernel Library Version 2022.2-Product Build 20220804 for Intel(R) 64 architecture applications
- Intel(R) MKL-DNN v2.7.3 (Git Hash 6dbeffbae1f23cbbeae17adb7b5b13f1f37c080e)
- OpenMP 201511 (a.k.a. OpenMP 4.5)
- LAPACK is enabled (usually provided by MKL)
- NNPACK is enabled
- CPU capability usage: AVX2
- CUDA Runtime 11.8
- NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90
- CuDNN 8.7
- Magma 2.6.1
- Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CUDA_VERSION=11.8, CUDNN_VERSION=8.7.0, CXX_COMPILER=/opt/rh/devtoolset-9/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=0 -fabi-version=11 -Wno-deprecated -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DUSE_FBGEMM -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wunused-local-typedefs -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-error=deprecated-declarations -Wno-stringop-overflow -Wno-psabi -Wno-error=pedantic -Wno-error=redundant-decls -Wno-error=old-style-cast -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Werror=cast-function-type -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_DISABLE_GPU_ASSERTS=ON, TORCH_VERSION=2.0.0, USE_CUDA=ON, USE_CUDNN=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF,
TorchVision: 0.15.1+cu118
OpenCV: 4.11.0
MMCV: 1.7.2
MMCV Compiler: GCC 9.3
MMCV CUDA Compiler: 11.8
MMRotate: 0.3.4+
2025-06-17 21:56:33,702 - mmrotate - INFO - Distributed training: False
2025-06-17 21:56:34,391 - mmrotate - INFO - Config:
dataset_type = 'DOTADataset'
data_root = '/home/DINO-MM/datasets/parcel_split/'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='LoadAnnotations', with_bbox=True),
dict(type='RResize', img_scale=(1024, 1024)),
dict(
type='RRandomFlip',
flip_ratio=[0.25, 0.25, 0.25],
direction=['horizontal', 'vertical', 'diagonal'],
version='le90'),
dict(
type='Normalize',
mean=[123.675, 116.28, 103.53],
std=[58.395, 57.12, 57.375],
to_rgb=True),
dict(type='Pad', size_divisor=32),
dict(type='DefaultFormatBundle'),
dict(type='Collect', keys=['img', 'gt_bboxes', 'gt_labels'])
]
test_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='MultiScaleFlipAug',
img_scale=(1120, 1120),
flip=False,
transforms=[
dict(type='RResize', img_scale=(1120, 1120)),
dict(
type='Normalize',
mean=[123.675, 116.28, 103.53],
std=[58.395, 57.12, 57.375],
to_rgb=True),
dict(type='Pad', size_divisor=14),
dict(type='DefaultFormatBundle'),
dict(type='Collect', keys=['img'])
])
]
data = dict(
samples_per_gpu=1,
workers_per_gpu=2,
train=dict(
type='DOTADataset',
ann_file='/home/DINO-MM/datasets/parcel_split/train/labels/',
img_prefix='/home/DINO-MM/datasets/parcel_split/train/images/',
pipeline=[
dict(type='LoadImageFromFile'),
dict(type='LoadAnnotations', with_bbox=True),
dict(type='RResize', img_scale=(1024, 1024)),
dict(
type='RRandomFlip',
flip_ratio=[0.25, 0.25, 0.25],
direction=['horizontal', 'vertical', 'diagonal'],
version='le90'),
dict(
type='Normalize',
mean=[123.675, 116.28, 103.53],
std=[58.395, 57.12, 57.375],
to_rgb=True),
dict(type='Pad', size_divisor=32),
dict(type='DefaultFormatBundle'),
dict(type='Collect', keys=['img', 'gt_bboxes', 'gt_labels'])
],
version='le90'),
val=dict(
type='DOTADataset',
ann_file='/home/DINO-MM/datasets/parcel_split/val/labels/',
img_prefix='/home/DINO-MM/datasets/parcel_split/val/images/',
pipeline=[
dict(type='LoadImageFromFile'),
dict(
type='MultiScaleFlipAug',
img_scale=(1120, 1120),
flip=False,
transforms=[
dict(type='RResize', img_scale=(1120, 1120)),
dict(
type='Normalize',
mean=[123.675, 116.28, 103.53],
std=[58.395, 57.12, 57.375],
to_rgb=True),
dict(type='Pad', size_divisor=14),
dict(type='DefaultFormatBundle'),
dict(type='Collect', keys=['img'])
])
],
version='le90'),
test=dict(
type='DOTADataset',
ann_file='/home/DINO-MM/datasets/parcel_split/test/labels/',
img_prefix='/home/DINO-MM/datasets/parcel_split/test/images/',
pipeline=[
dict(type='LoadImageFromFile'),
dict(
type='MultiScaleFlipAug',
img_scale=(1120, 1120),
flip=False,
transforms=[
dict(type='RResize', img_scale=(1120, 1120)),
dict(
type='Normalize',
mean=[123.675, 116.28, 103.53],
std=[58.395, 57.12, 57.375],
to_rgb=True),
dict(type='Pad', size_divisor=14),
dict(type='DefaultFormatBundle'),
dict(type='Collect', keys=['img'])
])
],
version='le90'))
evaluation = dict(interval=1, metric='mAP')
optimizer = dict(
type='AdamW',
lr=0.0001,
betas=(0.9, 0.999),
weight_decay=0.05,
paramwise_cfg=dict(
custom_keys=dict(
absolute_pos_embed=dict(decay_mult=0.0),
relative_position_bias_table=dict(decay_mult=0.0),
norm=dict(decay_mult=0.0))))
optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.3333333333333333,
step=[3, 6, 10])
runner = dict(type='EpochBasedRunner', max_epochs=12)
checkpoint_config = dict(interval=12)
log_config = dict(interval=50, hooks=[dict(type='TextLoggerHook')])
dist_params = dict(backend='nccl')
log_level = 'INFO'
load_from = None
resume_from = None
workflow = [('train', 1)]
opencv_num_threads = 0
mp_start_method = 'fork'
angle_version = 'le90'
model = dict(
type='OrientedRCNNCrop',
backbone=dict(
type='SwinBase',
pretrain_img_size=224,
embed_dims=192,
depths=[2, 2, 18, 2],
num_heads=[6, 12, 24, 48],
window_size=7,
patch_size=4,
mlp_ratio=4,
qkv_bias=True,
qk_scale=None,
drop_rate=0.0,
attn_drop_rate=0.0,
drop_path_rate=0.2,
patch_norm=True,
out_indices=(0, 1, 2, 3),
with_cp=False,
convert_weights=False,
frozen_stages=4,
init_cfg=dict(
type='Pretrained',
checkpoint=
'/root/autodl-tmp/swin_large_patch4_window7_224_22k_20220412-aeecf2aa.pth'
)),
neck=dict(
type='FPN',
in_channels=[192, 384, 768, 1536],
out_channels=256,
num_outs=5),
rpn_head=dict(
type='OrientedRPNHead',
in_channels=256,
feat_channels=256,
version='le90',
anchor_generator=dict(
type='AnchorGenerator',
scales=[8],
ratios=[0.5, 1.0, 2.0],
strides=[4, 8, 16, 32, 64]),
bbox_coder=dict(
type='MidpointOffsetCoder',
angle_range='le90',
target_means=[0.0, 0.0, 0.0, 0.0, 0.0, 0.0],
target_stds=[1.0, 1.0, 1.0, 1.0, 0.5, 0.5]),
loss_cls=dict(
type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
loss_bbox=dict(
type='SmoothL1Loss', beta=0.1111111111111111, loss_weight=1.0)),
roi_head=dict(
type='OrientedStandardRoIHead',
bbox_roi_extractor=dict(
type='RotatedSingleRoIExtractor',
roi_layer=dict(
type='RoIAlignRotated',
out_size=7,
sample_num=2,
clockwise=True),
out_channels=256,
featmap_strides=[4, 8, 16, 32]),
bbox_head=dict(
type='RotatedShared2FCBBoxHead',
in_channels=256,
fc_out_channels=1024,
roi_feat_size=7,
num_classes=1,
bbox_coder=dict(
type='DeltaXYWHAOBBoxCoder',
angle_range='le90',
norm_factor=None,
edge_swap=True,
proj_xy=True,
target_means=(0.0, 0.0, 0.0, 0.0, 0.0),
target_stds=(0.1, 0.1, 0.2, 0.2, 0.1)),
reg_class_agnostic=True,
loss_cls=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0),
loss_bbox=dict(type='SmoothL1Loss', beta=1.0, 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=0,
pos_weight=-1,
debug=False),
rpn_proposal=dict(
nms_pre=2000,
max_per_img=2000,
nms=dict(type='nms', iou_threshold=0.8),
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=False,
iou_calculator=dict(type='RBboxOverlaps2D'),
ignore_iof_thr=-1),
sampler=dict(
type='RRandomSampler',
num=512,
pos_fraction=0.25,
neg_pos_ub=-1,
add_gt_as_proposals=True),
pos_weight=-1,
debug=False)),
test_cfg=dict(
rpn=dict(
nms_pre=2000,
max_per_img=2000,
nms=dict(type='nms', iou_threshold=0.8),
min_bbox_size=0),
rcnn=dict(
nms_pre=2000,
min_bbox_size=0,
score_thr=0.05,
nms=dict(iou_thr=0.1),
max_per_img=2000)))
work_dir = 'out/20250617-swin'
auto_resume = False
gpu_ids = [0]
2025-06-17 21:56:34,392 - mmrotate - INFO - Set random seed to 1781076398, deterministic: False
2025-06-17 21:56:35,911 - mmrotate - INFO - load checkpoint from local path: /root/autodl-tmp/swin_large_patch4_window7_224_22k_20220412-aeecf2aa.pth
2025-06-17 21:56:37,564 - mmrotate - INFO - Start running, host: root@autodl-container-811348b9ea-3299e3bb, work_dir: /home/STAR-MM/out/20250617-swin
2025-06-17 21:56:37,564 - mmrotate - INFO - Hooks will be executed in the following order:
before_run:
(VERY_HIGH ) StepLrUpdaterHook
(NORMAL ) CheckpointHook
(LOW ) EvalHook
(VERY_LOW ) TextLoggerHook
before_train_epoch:
(VERY_HIGH ) StepLrUpdaterHook
(LOW ) IterTimerHook
(LOW ) EvalHook
(VERY_LOW ) TextLoggerHook
before_train_iter:
(VERY_HIGH ) StepLrUpdaterHook
(LOW ) IterTimerHook
(LOW ) EvalHook
after_train_iter:
(ABOVE_NORMAL) OptimizerHook
(NORMAL ) CheckpointHook
(LOW ) IterTimerHook
(LOW ) EvalHook
(VERY_LOW ) TextLoggerHook
after_train_epoch:
(NORMAL ) CheckpointHook
(LOW ) EvalHook
(VERY_LOW ) TextLoggerHook
before_val_epoch:
(LOW ) IterTimerHook
(VERY_LOW ) TextLoggerHook
before_val_iter:
(LOW ) IterTimerHook
after_val_iter:
(LOW ) IterTimerHook
after_val_epoch:
(VERY_LOW ) TextLoggerHook
after_run:
(VERY_LOW ) TextLoggerHook
2025-06-17 21:56:37,564 - mmrotate - INFO - workflow: [('train', 1)], max: 12 epochs
2025-06-17 21:56:37,564 - mmrotate - INFO - Checkpoints will be saved to /home/STAR-MM/out/20250617-swin by HardDiskBackend.
2025-06-17 21:56:47,245 - mmrotate - INFO - Epoch [1][50/355] lr: 3.987e-05, eta: 0:13:34, time: 0.193, data_time: 0.047, memory: 4334, loss_rpn_cls: 0.6111, loss_rpn_bbox: 0.8497, loss_cls: 0.6626, acc: 75.1953, loss_bbox: 0.6593, loss: 2.7827, grad_norm: 23.0183
2025-06-17 21:56:53,848 - mmrotate - INFO - Epoch [1][100/355] lr: 4.653e-05, eta: 0:11:17, time: 0.132, data_time: 0.004, memory: 5188, loss_rpn_cls: 0.5380, loss_rpn_bbox: 0.8290, loss_cls: 0.3755, acc: 83.8828, loss_bbox: 0.5324, loss: 2.2748, grad_norm: 10.2061
2025-06-17 21:57:00,421 - mmrotate - INFO - Epoch [1][150/355] lr: 5.320e-05, eta: 0:10:26, time: 0.131, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.4996, loss_rpn_bbox: 0.7845, loss_cls: 0.3603, acc: 85.9492, loss_bbox: 0.4156, loss: 2.0599, grad_norm: 9.2531
2025-06-17 21:57:06,881 - mmrotate - INFO - Epoch [1][200/355] lr: 5.987e-05, eta: 0:09:54, time: 0.129, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.4439, loss_rpn_bbox: 0.7703, loss_cls: 0.2966, acc: 88.1172, loss_bbox: 0.3139, loss: 1.8247, grad_norm: 8.9798
2025-06-17 21:57:13,516 - mmrotate - INFO - Epoch [1][250/355] lr: 6.653e-05, eta: 0:09:36, time: 0.133, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.4410, loss_rpn_bbox: 0.7610, loss_cls: 0.3021, acc: 87.9023, loss_bbox: 0.4314, loss: 1.9355, grad_norm: 9.4459
2025-06-17 21:57:20,003 - mmrotate - INFO - Epoch [1][300/355] lr: 7.320e-05, eta: 0:09:20, time: 0.130, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.4021, loss_rpn_bbox: 0.7577, loss_cls: 0.3051, acc: 87.6133, loss_bbox: 0.5329, loss: 1.9979, grad_norm: 8.1019
2025-06-17 21:57:26,314 - mmrotate - INFO - Epoch [1][350/355] lr: 7.987e-05, eta: 0:09:04, time: 0.126, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.3931, loss_rpn_bbox: 0.7704, loss_cls: 0.3091, acc: 86.9102, loss_bbox: 0.6064, loss: 2.0790, grad_norm: 7.9707
2025-06-17 21:58:39,402 - mmrotate - INFO -
+--------+-------+-------+--------+-------+
| class | gts | dets | recall | ap |
+--------+-------+-------+--------+-------+
| parcel | 15205 | 19452 | 0.101 | 0.053 |
+--------+-------+-------+--------+-------+
| mAP | | | | 0.053 |
+--------+-------+-------+--------+-------+
2025-06-17 21:58:39,459 - mmrotate - INFO - Exp name: oriented_rcnn_swin-l_fpn_1x_star_le90.py
2025-06-17 21:58:39,459 - mmrotate - INFO - Epoch(val) [1][103] mAP: 0.0530
2025-06-17 21:58:48,266 - mmrotate - INFO - Epoch [2][50/355] lr: 8.720e-05, eta: 0:09:07, time: 0.176, data_time: 0.048, memory: 6006, loss_rpn_cls: 0.3685, loss_rpn_bbox: 0.7464, loss_cls: 0.3703, acc: 83.9375, loss_bbox: 0.7484, loss: 2.2337, grad_norm: 7.9442
2025-06-17 21:58:54,836 - mmrotate - INFO - Epoch [2][100/355] lr: 9.387e-05, eta: 0:08:56, time: 0.131, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.3126, loss_rpn_bbox: 0.6935, loss_cls: 0.3542, acc: 84.5859, loss_bbox: 0.6903, loss: 2.0505, grad_norm: 8.1627
2025-06-17 21:59:01,273 - mmrotate - INFO - Epoch [2][150/355] lr: 1.000e-04, eta: 0:08:44, time: 0.129, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.2981, loss_rpn_bbox: 0.6814, loss_cls: 0.3608, acc: 83.8633, loss_bbox: 0.6556, loss: 1.9960, grad_norm: 8.2547
2025-06-17 21:59:07,771 - mmrotate - INFO - Epoch [2][200/355] lr: 1.000e-04, eta: 0:08:34, time: 0.130, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.2712, loss_rpn_bbox: 0.6370, loss_cls: 0.3674, acc: 84.0508, loss_bbox: 0.5900, loss: 1.8657, grad_norm: 7.8070
2025-06-17 21:59:14,159 - mmrotate - INFO - Epoch [2][250/355] lr: 1.000e-04, eta: 0:08:23, time: 0.128, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.2588, loss_rpn_bbox: 0.6577, loss_cls: 0.3491, acc: 85.1055, loss_bbox: 0.6638, loss: 1.9294, grad_norm: 8.6752
2025-06-17 21:59:20,590 - mmrotate - INFO - Epoch [2][300/355] lr: 1.000e-04, eta: 0:08:14, time: 0.129, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.2520, loss_rpn_bbox: 0.6149, loss_cls: 0.3617, acc: 84.2578, loss_bbox: 0.6414, loss: 1.8700, grad_norm: 7.2217
2025-06-17 21:59:27,121 - mmrotate - INFO - Epoch [2][350/355] lr: 1.000e-04, eta: 0:08:06, time: 0.131, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.2348, loss_rpn_bbox: 0.6690, loss_cls: 0.3821, acc: 82.7852, loss_bbox: 0.7649, loss: 2.0508, grad_norm: 7.7170
2025-06-17 22:00:43,318 - mmrotate - INFO -
+--------+-------+-------+--------+-------+
| class | gts | dets | recall | ap |
+--------+-------+-------+--------+-------+
| parcel | 15205 | 18905 | 0.172 | 0.050 |
+--------+-------+-------+--------+-------+
| mAP | | | | 0.050 |
+--------+-------+-------+--------+-------+
2025-06-17 22:00:43,371 - mmrotate - INFO - Exp name: oriented_rcnn_swin-l_fpn_1x_star_le90.py
2025-06-17 22:00:43,372 - mmrotate - INFO - Epoch(val) [2][103] mAP: 0.0498
2025-06-17 22:00:51,966 - mmrotate - INFO - Epoch [3][50/355] lr: 1.000e-04, eta: 0:08:03, time: 0.171, data_time: 0.047, memory: 6006, loss_rpn_cls: 0.2306, loss_rpn_bbox: 0.5893, loss_cls: 0.3781, acc: 83.4570, loss_bbox: 0.7267, loss: 1.9247, grad_norm: 8.1029
2025-06-17 22:00:58,585 - mmrotate - INFO - Epoch [3][100/355] lr: 1.000e-04, eta: 0:07:55, time: 0.132, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.2347, loss_rpn_bbox: 0.6289, loss_cls: 0.3705, acc: 83.7383, loss_bbox: 0.6642, loss: 1.8984, grad_norm: 7.5750
2025-06-17 22:01:05,137 - mmrotate - INFO - Epoch [3][150/355] lr: 1.000e-04, eta: 0:07:46, time: 0.131, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.2263, loss_rpn_bbox: 0.6114, loss_cls: 0.3699, acc: 83.9805, loss_bbox: 0.6965, loss: 1.9041, grad_norm: 7.7494
2025-06-17 22:01:11,452 - mmrotate - INFO - Epoch [3][200/355] lr: 1.000e-04, eta: 0:07:38, time: 0.126, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.2199, loss_rpn_bbox: 0.6036, loss_cls: 0.3750, acc: 83.2539, loss_bbox: 0.6771, loss: 1.8756, grad_norm: 7.7833
2025-06-17 22:01:17,973 - mmrotate - INFO - Epoch [3][250/355] lr: 1.000e-04, eta: 0:07:30, time: 0.130, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.2061, loss_rpn_bbox: 0.5637, loss_cls: 0.3689, acc: 83.7031, loss_bbox: 0.6346, loss: 1.7733, grad_norm: 7.9309
2025-06-17 22:01:24,561 - mmrotate - INFO - Epoch [3][300/355] lr: 1.000e-04, eta: 0:07:22, time: 0.132, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.2031, loss_rpn_bbox: 0.5943, loss_cls: 0.3812, acc: 83.0234, loss_bbox: 0.6125, loss: 1.7911, grad_norm: 7.6124
2025-06-17 22:01:30,922 - mmrotate - INFO - Epoch [3][350/355] lr: 1.000e-04, eta: 0:07:14, time: 0.127, data_time: 0.003, memory: 6006, loss_rpn_cls: 0.2101, loss_rpn_bbox: 0.5982, loss_cls: 0.3722, acc: 83.4180, loss_bbox: 0.6209, loss: 1.8013, grad_norm: 8.1166
2025-06-17 22:02:44,641 - mmrotate - INFO -
+--------+-------+-------+--------+-------+
| class | gts | dets | recall | ap |
+--------+-------+-------+--------+-------+
| parcel | 15205 | 19710 | 0.201 | 0.074 |
+--------+-------+-------+--------+-------+
| mAP | | | | 0.074 |
+--------+-------+-------+--------+-------+
2025-06-17 22:02:44,694 - mmrotate - INFO - Exp name: oriented_rcnn_swin-l_fpn_1x_star_le90.py
2025-06-17 22:02:44,695 - mmrotate - INFO - Epoch(val) [3][103] mAP: 0.0740
2025-06-17 22:02:53,539 - mmrotate - INFO - Epoch [4][50/355] lr: 1.000e-05, eta: 0:07:10, time: 0.177, data_time: 0.050, memory: 6006, loss_rpn_cls: 0.2088, loss_rpn_bbox: 0.5613, loss_cls: 0.3729, acc: 83.5469, loss_bbox: 0.5668, loss: 1.7098, grad_norm: 6.6645
2025-06-17 22:03:00,147 - mmrotate - INFO - Epoch [4][100/355] lr: 1.000e-05, eta: 0:07:03, time: 0.132, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1942, loss_rpn_bbox: 0.5295, loss_cls: 0.3617, acc: 83.8867, loss_bbox: 0.5345, loss: 1.6199, grad_norm: 6.5634
2025-06-17 22:03:06,776 - mmrotate - INFO - Epoch [4][150/355] lr: 1.000e-05, eta: 0:06:55, time: 0.133, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1900, loss_rpn_bbox: 0.5269, loss_cls: 0.3456, acc: 84.6875, loss_bbox: 0.5137, loss: 1.5762, grad_norm: 6.3105
2025-06-17 22:03:13,302 - mmrotate - INFO - Epoch [4][200/355] lr: 1.000e-05, eta: 0:06:48, time: 0.131, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1815, loss_rpn_bbox: 0.5211, loss_cls: 0.3507, acc: 84.4531, loss_bbox: 0.5431, loss: 1.5964, grad_norm: 6.9466
2025-06-17 22:03:19,740 - mmrotate - INFO - Epoch [4][250/355] lr: 1.000e-05, eta: 0:06:40, time: 0.129, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1949, loss_rpn_bbox: 0.5622, loss_cls: 0.3609, acc: 84.0938, loss_bbox: 0.5794, loss: 1.6975, grad_norm: 6.9612
2025-06-17 22:03:26,181 - mmrotate - INFO - Epoch [4][300/355] lr: 1.000e-05, eta: 0:06:33, time: 0.129, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1823, loss_rpn_bbox: 0.5422, loss_cls: 0.3508, acc: 84.6562, loss_bbox: 0.5499, loss: 1.6253, grad_norm: 6.4031
2025-06-17 22:03:32,570 - mmrotate - INFO - Epoch [4][350/355] lr: 1.000e-05, eta: 0:06:25, time: 0.128, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1915, loss_rpn_bbox: 0.5485, loss_cls: 0.3607, acc: 84.1797, loss_bbox: 0.5584, loss: 1.6591, grad_norm: 6.8228
2025-06-17 22:04:50,205 - mmrotate - INFO -
+--------+-------+-------+--------+-------+
| class | gts | dets | recall | ap |
+--------+-------+-------+--------+-------+
| parcel | 15205 | 20211 | 0.209 | 0.082 |
+--------+-------+-------+--------+-------+
| mAP | | | | 0.082 |
+--------+-------+-------+--------+-------+
2025-06-17 22:04:50,259 - mmrotate - INFO - Exp name: oriented_rcnn_swin-l_fpn_1x_star_le90.py
2025-06-17 22:04:50,259 - mmrotate - INFO - Epoch(val) [4][103] mAP: 0.0820
2025-06-17 22:04:58,982 - mmrotate - INFO - Epoch [5][50/355] lr: 1.000e-05, eta: 0:06:20, time: 0.174, data_time: 0.048, memory: 6006, loss_rpn_cls: 0.1895, loss_rpn_bbox: 0.5371, loss_cls: 0.3615, acc: 84.0039, loss_bbox: 0.5495, loss: 1.6376, grad_norm: 6.3270
2025-06-17 22:05:05,526 - mmrotate - INFO - Epoch [5][100/355] lr: 1.000e-05, eta: 0:06:13, time: 0.131, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1767, loss_rpn_bbox: 0.5581, loss_cls: 0.3471, acc: 84.9492, loss_bbox: 0.5307, loss: 1.6126, grad_norm: 7.0356
2025-06-17 22:05:12,023 - mmrotate - INFO - Epoch [5][150/355] lr: 1.000e-05, eta: 0:06:05, time: 0.130, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1782, loss_rpn_bbox: 0.5790, loss_cls: 0.3591, acc: 84.1875, loss_bbox: 0.5424, loss: 1.6587, grad_norm: 7.3820
2025-06-17 22:05:18,622 - mmrotate - INFO - Epoch [5][200/355] lr: 1.000e-05, eta: 0:05:58, time: 0.132, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1853, loss_rpn_bbox: 0.5200, loss_cls: 0.3609, acc: 84.1562, loss_bbox: 0.5309, loss: 1.5970, grad_norm: 6.3677
2025-06-17 22:05:24,997 - mmrotate - INFO - Epoch [5][250/355] lr: 1.000e-05, eta: 0:05:51, time: 0.127, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1913, loss_rpn_bbox: 0.4979, loss_cls: 0.3583, acc: 84.3711, loss_bbox: 0.5206, loss: 1.5681, grad_norm: 6.2940
2025-06-17 22:05:31,392 - mmrotate - INFO - Epoch [5][300/355] lr: 1.000e-05, eta: 0:05:43, time: 0.128, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1935, loss_rpn_bbox: 0.5142, loss_cls: 0.3520, acc: 84.5508, loss_bbox: 0.5161, loss: 1.5758, grad_norm: 6.6237
2025-06-17 22:05:37,915 - mmrotate - INFO - Epoch [5][350/355] lr: 1.000e-05, eta: 0:05:36, time: 0.130, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1875, loss_rpn_bbox: 0.5130, loss_cls: 0.3442, acc: 84.8477, loss_bbox: 0.4978, loss: 1.5425, grad_norm: 6.6980
2025-06-17 22:06:58,184 - mmrotate - INFO -
+--------+-------+-------+--------+-------+
| class | gts | dets | recall | ap |
+--------+-------+-------+--------+-------+
| parcel | 15205 | 20989 | 0.211 | 0.078 |
+--------+-------+-------+--------+-------+
| mAP | | | | 0.078 |
+--------+-------+-------+--------+-------+
2025-06-17 22:06:58,244 - mmrotate - INFO - Exp name: oriented_rcnn_swin-l_fpn_1x_star_le90.py
2025-06-17 22:06:58,244 - mmrotate - INFO - Epoch(val) [5][103] mAP: 0.0778
2025-06-17 22:07:07,031 - mmrotate - INFO - Epoch [6][50/355] lr: 1.000e-05, eta: 0:05:31, time: 0.175, data_time: 0.047, memory: 6006, loss_rpn_cls: 0.1820, loss_rpn_bbox: 0.5113, loss_cls: 0.3678, acc: 83.7969, loss_bbox: 0.5467, loss: 1.6080, grad_norm: 6.5147
2025-06-17 22:07:13,561 - mmrotate - INFO - Epoch [6][100/355] lr: 1.000e-05, eta: 0:05:23, time: 0.131, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1796, loss_rpn_bbox: 0.5215, loss_cls: 0.3581, acc: 84.1680, loss_bbox: 0.5287, loss: 1.5880, grad_norm: 6.7049
2025-06-17 22:07:20,029 - mmrotate - INFO - Epoch [6][150/355] lr: 1.000e-05, eta: 0:05:16, time: 0.129, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1787, loss_rpn_bbox: 0.5664, loss_cls: 0.3530, acc: 84.3828, loss_bbox: 0.5640, loss: 1.6620, grad_norm: 6.8010
2025-06-17 22:07:26,357 - mmrotate - INFO - Epoch [6][200/355] lr: 1.000e-05, eta: 0:05:09, time: 0.127, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1856, loss_rpn_bbox: 0.4961, loss_cls: 0.3547, acc: 84.2656, loss_bbox: 0.5046, loss: 1.5410, grad_norm: 7.1144
2025-06-17 22:07:32,758 - mmrotate - INFO - Epoch [6][250/355] lr: 1.000e-05, eta: 0:05:02, time: 0.128, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1822, loss_rpn_bbox: 0.5413, loss_cls: 0.3545, acc: 84.4219, loss_bbox: 0.5497, loss: 1.6277, grad_norm: 6.9885
2025-06-17 22:07:39,135 - mmrotate - INFO - Epoch [6][300/355] lr: 1.000e-05, eta: 0:04:55, time: 0.128, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1837, loss_rpn_bbox: 0.5008, loss_cls: 0.3451, acc: 84.8945, loss_bbox: 0.5160, loss: 1.5457, grad_norm: 6.6634
2025-06-17 22:07:45,467 - mmrotate - INFO - Epoch [6][350/355] lr: 1.000e-05, eta: 0:04:47, time: 0.127, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1796, loss_rpn_bbox: 0.5608, loss_cls: 0.3525, acc: 84.4805, loss_bbox: 0.5206, loss: 1.6134, grad_norm: 7.4153
2025-06-17 22:09:02,455 - mmrotate - INFO -
+--------+-------+-------+--------+-------+
| class | gts | dets | recall | ap |
+--------+-------+-------+--------+-------+
| parcel | 15205 | 20865 | 0.216 | 0.079 |
+--------+-------+-------+--------+-------+
| mAP | | | | 0.079 |
+--------+-------+-------+--------+-------+
2025-06-17 22:09:02,510 - mmrotate - INFO - Exp name: oriented_rcnn_swin-l_fpn_1x_star_le90.py
2025-06-17 22:09:02,510 - mmrotate - INFO - Epoch(val) [6][103] mAP: 0.0787
2025-06-17 22:09:11,085 - mmrotate - INFO - Epoch [7][50/355] lr: 1.000e-06, eta: 0:04:41, time: 0.171, data_time: 0.047, memory: 6006, loss_rpn_cls: 0.1669, loss_rpn_bbox: 0.4911, loss_cls: 0.3407, acc: 84.7852, loss_bbox: 0.5037, loss: 1.5023, grad_norm: 6.6266
2025-06-17 22:09:17,553 - mmrotate - INFO - Epoch [7][100/355] lr: 1.000e-06, eta: 0:04:34, time: 0.129, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1759, loss_rpn_bbox: 0.5562, loss_cls: 0.3485, acc: 84.4648, loss_bbox: 0.5219, loss: 1.6026, grad_norm: 7.1460
2025-06-17 22:09:24,042 - mmrotate - INFO - Epoch [7][150/355] lr: 1.000e-06, eta: 0:04:27, time: 0.130, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1803, loss_rpn_bbox: 0.5121, loss_cls: 0.3443, acc: 84.6758, loss_bbox: 0.5240, loss: 1.5607, grad_norm: 6.7061
2025-06-17 22:09:30,553 - mmrotate - INFO - Epoch [7][200/355] lr: 1.000e-06, eta: 0:04:20, time: 0.130, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1887, loss_rpn_bbox: 0.5509, loss_cls: 0.3493, acc: 84.7383, loss_bbox: 0.5221, loss: 1.6110, grad_norm: 6.6691
2025-06-17 22:09:37,231 - mmrotate - INFO - Epoch [7][250/355] lr: 1.000e-06, eta: 0:04:13, time: 0.134, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1932, loss_rpn_bbox: 0.5093, loss_cls: 0.3593, acc: 83.7578, loss_bbox: 0.5174, loss: 1.5793, grad_norm: 6.4679
2025-06-17 22:09:43,660 - mmrotate - INFO - Epoch [7][300/355] lr: 1.000e-06, eta: 0:04:06, time: 0.129, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1734, loss_rpn_bbox: 0.4929, loss_cls: 0.3488, acc: 84.5781, loss_bbox: 0.4911, loss: 1.5063, grad_norm: 6.9015
2025-06-17 22:09:50,003 - mmrotate - INFO - Epoch [7][350/355] lr: 1.000e-06, eta: 0:03:59, time: 0.127, data_time: 0.004, memory: 6006, loss_rpn_cls: 0.1752, loss_rpn_bbox: 0.5304, loss_cls: 0.3534, acc: 84.2812, loss_bbox: 0.5139, loss: 1.5729, grad_norm: 6.3996
2025-06-17 22:11:09,997 - mmrotate - INFO -
+--------+-------+-------+--------+-------+
| class | gts | dets | recall | ap |
+--------+-------+-------+--------+-------+
| parcel | 15205 | 20906 | 0.215 | 0.092 |
+--------+-------+-------+--------+-------+
| mAP | | | | 0.092 |
+--------+-------+-------+--------+-------+
2025-06-17 22:11:10,052 - mmrotate - INFO - Exp name: oriented_rcnn_swin-l_fpn_1x_star_le90.py
2025-06-17 22:11:10,053 - mmrotate - INFO - Epoch(val) [7][103] mAP: 0.0923
2025-06-17 22:11:18,669 - mmrotate - INFO - Epoch [8][50/355] lr: 1.000e-06, eta: 0:03:53, time: 0.172, data_time: 0.047, memory: 6006, loss_rpn_cls: 0.1883, loss_rpn_bbox: 0.5036, loss_cls: 0.3548, acc: 84.1328, loss_bbox: 0.5393, loss: 1.5860, grad_norm: 6.1960