[Fix] Fix UT and remove delete mmcv ops. (open-mmlab#8623)

* Remove get_root_logger * Fix UT * Update
BIGWangYuDong · Aug 26, 2022 · ceae85b · ceae85b
1 parent 444654a
commit ceae85b
Show file tree

Hide file tree

Showing 34 changed files with 63 additions and 93 deletions.
diff --git a/docs/en/tutorials/customize_runtime.md b/docs/en/tutorials/customize_runtime.md
@@ -84,7 +84,6 @@ The users can do those fine-grained parameter tuning through customizing optimiz
 from mmcv.utils import build_from_cfg
 
 from mmcv.runner.optimizer import OPTIMIZER_BUILDERS, OPTIMIZERS
-from mmdet.utils import get_root_logger
 from .my_optimizer import MyOptimizer
 
 

diff --git a/mmdet/engine/optimizers/layer_decay_optimizer_constructor.py b/mmdet/engine/optimizers/layer_decay_optimizer_constructor.py
@@ -4,10 +4,10 @@
 
 import torch.nn as nn
 from mmengine.dist import get_dist_info
+from mmengine.logging import MMLogger
 from mmengine.optim import DefaultOptimWrapperConstructor
 
 from mmdet.registry import OPTIM_WRAPPER_CONSTRUCTORS
-from mmdet.utils import get_root_logger
 
 
 def get_layer_id_for_convnext(var_name, max_layer_id):
@@ -95,7 +95,7 @@ def add_params(self, params: List[dict], module: nn.Module,
                 in place.
             module (nn.Module): The module to be added.
         """
-        logger = get_root_logger()
+        logger = MMLogger.get_current_instance()
 
         parameter_groups = {}
         logger.info(f'self.paramwise_cfg is {self.paramwise_cfg}')

diff --git a/mmdet/models/backbones/detectors_resnet.py b/mmdet/models/backbones/detectors_resnet.py
@@ -2,13 +2,13 @@
 import torch.nn as nn
 import torch.utils.checkpoint as cp
 from mmcv.cnn import build_conv_layer, build_norm_layer
+from mmengine.logging import MMLogger
 from mmengine.model import Sequential
 from mmengine.model.utils import constant_init, kaiming_init
 from mmengine.runner.checkpoint import load_checkpoint
 from torch.nn.modules.batchnorm import _BatchNorm
 
 from mmdet.registry import MODELS
-from mmdet.utils import get_root_logger
 from .resnet import BasicBlock
 from .resnet import Bottleneck as _Bottleneck
 from .resnet import ResNet
@@ -299,7 +299,7 @@ def init_weights(self):
         # super(DetectoRS_ResNet, self).init_weights()
 
         if isinstance(self.pretrained, str):
-            logger = get_root_logger()
+            logger = MMLogger.get_current_instance()
             load_checkpoint(self, self.pretrained, strict=False, logger=logger)
         elif self.pretrained is None:
             for m in self.modules():

diff --git a/mmdet/models/backbones/pvt.py b/mmdet/models/backbones/pvt.py
@@ -10,14 +10,14 @@
 from mmcv.cnn import Conv2d, build_activation_layer, build_norm_layer
 from mmcv.cnn.bricks.drop import build_dropout
 from mmcv.cnn.bricks.transformer import MultiheadAttention
+from mmengine.logging import MMLogger
 from mmengine.model import BaseModule, ModuleList, Sequential
 from mmengine.model.utils import (constant_init, normal_init, trunc_normal_,
                                   trunc_normal_init)
 from mmengine.runner.checkpoint import CheckpointLoader, load_state_dict
 from torch.nn.modules.utils import _pair as to_2tuple
 
 from mmdet.registry import MODELS
-from ...utils import get_root_logger
 from ..layers import PatchEmbed, nchw_to_nlc, nlc_to_nchw
 
 
@@ -522,7 +522,7 @@ def __init__(self,
             cur += num_layer
 
     def init_weights(self):
-        logger = get_root_logger()
+        logger = MMLogger.get_current_instance()
         if self.init_cfg is None:
             logger.warn(f'No pre-trained weights for '
                         f'{self.__class__.__name__}, '

diff --git a/mmdet/models/backbones/swin.py b/mmdet/models/backbones/swin.py
@@ -9,14 +9,14 @@
 import torch.utils.checkpoint as cp
 from mmcv.cnn import build_norm_layer
 from mmcv.cnn.bricks.transformer import FFN, build_dropout
+from mmengine.logging import MMLogger
 from mmengine.model import BaseModule, ModuleList
 from mmengine.model.utils import (constant_init, trunc_normal_,
                                   trunc_normal_init)
 from mmengine.runner.checkpoint import CheckpointLoader
 from mmengine.utils import to_2tuple
 
 from mmdet.registry import MODELS
-from ...utils import get_root_logger
 from ..layers import PatchEmbed, PatchMerging
 
 
@@ -669,7 +669,7 @@ def _freeze_stages(self):
                 param.requires_grad = False
 
     def init_weights(self):
-        logger = get_root_logger()
+        logger = MMLogger.get_current_instance()
         if self.init_cfg is None:
             logger.warn(f'No pre-trained weights for '
                         f'{self.__class__.__name__}, '

diff --git a/mmdet/models/losses/accuracy.py b/mmdet/models/losses/accuracy.py
@@ -1,9 +1,7 @@
 # Copyright (c) OpenMMLab. All rights reserved.
-import mmcv
 import torch.nn as nn
 
 
-@mmcv.jit(coderize=True)
 def accuracy(pred, target, topk=1, thresh=None):
     """Calculate accuracy according to the prediction and target.
 

diff --git a/mmdet/models/losses/ae_loss.py b/mmdet/models/losses/ae_loss.py
@@ -1,13 +1,11 @@
 # Copyright (c) OpenMMLab. All rights reserved.
-import mmcv
 import torch
 import torch.nn as nn
 import torch.nn.functional as F
 
 from mmdet.registry import MODELS
 
 
-@mmcv.jit(derivate=True, coderize=True)
 def ae_loss_per_image(tl_preds, br_preds, match):
     """Associative Embedding Loss in one image.
 

diff --git a/mmdet/models/losses/balanced_l1_loss.py b/mmdet/models/losses/balanced_l1_loss.py
@@ -1,5 +1,4 @@
 # Copyright (c) OpenMMLab. All rights reserved.
-import mmcv
 import numpy as np
 import torch
 import torch.nn as nn
@@ -8,7 +7,6 @@
 from .utils import weighted_loss
 
 
-@mmcv.jit(derivate=True, coderize=True)
 @weighted_loss
 def balanced_l1_loss(pred,
                      target,

diff --git a/mmdet/models/losses/gfocal_loss.py b/mmdet/models/losses/gfocal_loss.py
@@ -1,13 +1,11 @@
 # Copyright (c) OpenMMLab. All rights reserved.
-import mmcv
 import torch.nn as nn
 import torch.nn.functional as F
 
 from mmdet.registry import MODELS
 from .utils import weighted_loss
 
 
-@mmcv.jit(derivate=True, coderize=True)
 @weighted_loss
 def quality_focal_loss(pred, target, beta=2.0):
     r"""Quality Focal Loss (QFL) is from `Generalized Focal Loss: Learning
@@ -98,7 +96,6 @@ def quality_focal_loss_with_prob(pred, target, beta=2.0):
     return loss
 
 
-@mmcv.jit(derivate=True, coderize=True)
 @weighted_loss
 def distribution_focal_loss(pred, label):
     r"""Distribution Focal Loss (DFL) is from `Generalized Focal Loss: Learning

diff --git a/mmdet/models/losses/iou_loss.py b/mmdet/models/losses/iou_loss.py
@@ -2,7 +2,6 @@
 import math
 import warnings
 
-import mmcv
 import torch
 import torch.nn as nn
 
@@ -11,7 +10,6 @@
 from .utils import weighted_loss
 
 
-@mmcv.jit(derivate=True, coderize=True)
 @weighted_loss
 def iou_loss(pred, target, linear=False, mode='log', eps=1e-6):
     """IoU loss.
@@ -50,7 +48,6 @@ def iou_loss(pred, target, linear=False, mode='log', eps=1e-6):
     return loss
 
 
-@mmcv.jit(derivate=True, coderize=True)
 @weighted_loss
 def bounded_iou_loss(pred, target, beta=0.2, eps=1e-3):
     """BIoULoss.
@@ -97,7 +94,6 @@ def bounded_iou_loss(pred, target, beta=0.2, eps=1e-3):
     return loss
 
 
-@mmcv.jit(derivate=True, coderize=True)
 @weighted_loss
 def giou_loss(pred, target, eps=1e-7):
     r"""`Generalized Intersection over Union: A Metric and A Loss for Bounding
@@ -117,7 +113,6 @@ def giou_loss(pred, target, eps=1e-7):
     return loss
 
 
-@mmcv.jit(derivate=True, coderize=True)
 @weighted_loss
 def diou_loss(pred, target, eps=1e-7):
     r"""`Implementation of Distance-IoU Loss: Faster and Better
@@ -172,7 +167,6 @@ def diou_loss(pred, target, eps=1e-7):
     return loss
 
 
-@mmcv.jit(derivate=True, coderize=True)
 @weighted_loss
 def ciou_loss(pred, target, eps=1e-7):
     r"""`Implementation of paper `Enhancing Geometric Factors into

diff --git a/mmdet/models/losses/kd_loss.py b/mmdet/models/losses/kd_loss.py
@@ -1,13 +1,11 @@
 # Copyright (c) OpenMMLab. All rights reserved.
-import mmcv
 import torch.nn as nn
 import torch.nn.functional as F
 
 from mmdet.registry import MODELS
 from .utils import weighted_loss
 
 
-@mmcv.jit(derivate=True, coderize=True)
 @weighted_loss
 def knowledge_distillation_kl_div_loss(pred,
                                        soft_label,

diff --git a/mmdet/models/losses/pisa_loss.py b/mmdet/models/losses/pisa_loss.py
@@ -1,7 +1,6 @@
 # Copyright (c) OpenMMLab. All rights reserved.
 from typing import List, Optional, Tuple
 
-import mmcv
 import torch
 import torch.nn as nn
 from torch import Tensor
@@ -11,7 +10,6 @@
 from ..task_modules.samplers import SamplingResult
 
 
-@mmcv.jit(derivate=True, coderize=True)
 def isr_p(cls_score: Tensor,
           bbox_pred: Tensor,
           bbox_targets: Tuple[Tensor],
@@ -125,7 +123,6 @@ def isr_p(cls_score: Tensor,
     return bbox_targets
 
 
-@mmcv.jit(derivate=True, coderize=True)
 def carl_loss(cls_score: Tensor,
               labels: Tensor,
               bbox_pred: Tensor,

diff --git a/mmdet/models/losses/smooth_l1_loss.py b/mmdet/models/losses/smooth_l1_loss.py
@@ -1,13 +1,11 @@
 # Copyright (c) OpenMMLab. All rights reserved.
-import mmcv
 import torch
 import torch.nn as nn
 
 from mmdet.registry import MODELS
 from .utils import weighted_loss
 
 
-@mmcv.jit(derivate=True, coderize=True)
 @weighted_loss
 def smooth_l1_loss(pred, target, beta=1.0):
     """Smooth L1 loss.
@@ -32,7 +30,6 @@ def smooth_l1_loss(pred, target, beta=1.0):
     return loss
 
 
-@mmcv.jit(derivate=True, coderize=True)
 @weighted_loss
 def l1_loss(pred, target):
     """L1 loss.

diff --git a/mmdet/models/losses/utils.py b/mmdet/models/losses/utils.py
@@ -1,7 +1,6 @@
 # Copyright (c) OpenMMLab. All rights reserved.
 import functools
 
-import mmcv
 import torch
 import torch.nn.functional as F
 
@@ -26,7 +25,6 @@ def reduce_loss(loss, reduction):
         return loss.sum()
 
 
-@mmcv.jit(derivate=True, coderize=True)
 def weight_reduce_loss(loss, weight=None, reduction='mean', avg_factor=None):
     """Apply element-wise weight and reduce loss.
 

diff --git a/mmdet/models/losses/varifocal_loss.py b/mmdet/models/losses/varifocal_loss.py
@@ -1,13 +1,11 @@
 # Copyright (c) OpenMMLab. All rights reserved.
-import mmcv
 import torch.nn as nn
 import torch.nn.functional as F
 
 from mmdet.registry import MODELS
 from .utils import weight_reduce_loss
 
 
-@mmcv.jit(derivate=True, coderize=True)
 def varifocal_loss(pred,
                    target,
                    weight=None,

diff --git a/mmdet/models/task_modules/coders/bucketing_bbox_coder.py b/mmdet/models/task_modules/coders/bucketing_bbox_coder.py
@@ -1,5 +1,4 @@
 # Copyright (c) OpenMMLab. All rights reserved.
-import mmcv
 import numpy as np
 import torch
 import torch.nn.functional as F
@@ -92,7 +91,6 @@ def decode(self, bboxes, pred_bboxes, max_shape=None):
         return decoded_bboxes
 
 
-@mmcv.jit(coderize=True)
 def generat_buckets(proposals, num_buckets, scale_factor=1.0):
     """Generate buckets w.r.t bucket number and scale factor of proposals.
 
@@ -141,7 +139,6 @@ def generat_buckets(proposals, num_buckets, scale_factor=1.0):
     return bucket_w, bucket_h, l_buckets, r_buckets, t_buckets, d_buckets
 
 
-@mmcv.jit(coderize=True)
 def bbox2bucket(proposals,
                 gt,
                 num_buckets,
@@ -265,7 +262,6 @@ def bbox2bucket(proposals,
     return offsets, offsets_weights, bucket_labels, bucket_cls_weights
 
 
-@mmcv.jit(coderize=True)
 def bucket2bbox(proposals,
                 cls_preds,
                 offset_preds,

diff --git a/mmdet/models/task_modules/coders/delta_xywh_bbox_coder.py b/mmdet/models/task_modules/coders/delta_xywh_bbox_coder.py
@@ -1,7 +1,6 @@
 # Copyright (c) OpenMMLab. All rights reserved.
 import warnings
 
-import mmcv
 import numpy as np
 import torch
 
@@ -114,7 +113,6 @@ def decode(self,
         return decoded_bboxes
 
 
-@mmcv.jit(coderize=True)
 def bbox2delta(proposals, gt, means=(0., 0., 0., 0.), stds=(1., 1., 1., 1.)):
     """Compute deltas of proposals w.r.t. gt.
 
@@ -160,7 +158,6 @@ def bbox2delta(proposals, gt, means=(0., 0., 0., 0.), stds=(1., 1., 1., 1.)):
     return deltas
 
 
-@mmcv.jit(coderize=True)
 def delta2bbox(rois,
                deltas,
                means=(0., 0., 0., 0.),

diff --git a/mmdet/models/task_modules/coders/legacy_delta_xywh_bbox_coder.py b/mmdet/models/task_modules/coders/legacy_delta_xywh_bbox_coder.py
@@ -1,5 +1,4 @@
 # Copyright (c) OpenMMLab. All rights reserved.
-import mmcv
 import numpy as np
 import torch
 
@@ -81,7 +80,6 @@ def decode(self,
         return decoded_bboxes
 
 
-@mmcv.jit(coderize=True)
 def legacy_bbox2delta(proposals,
                       gt,
                       means=(0., 0., 0., 0.),
@@ -130,7 +128,6 @@ def legacy_bbox2delta(proposals,
     return deltas
 
 
-@mmcv.jit(coderize=True)
 def legacy_delta2bbox(rois,
                       deltas,
                       means=(0., 0., 0., 0.),

diff --git a/mmdet/models/task_modules/coders/tblr_bbox_coder.py b/mmdet/models/task_modules/coders/tblr_bbox_coder.py
@@ -1,5 +1,4 @@
 # Copyright (c) OpenMMLab. All rights reserved.
-import mmcv
 import torch
 
 from mmdet.registry import TASK_UTILS
@@ -73,7 +72,6 @@ def decode(self, bboxes, pred_bboxes, max_shape=None):
         return decoded_bboxes
 
 
-@mmcv.jit(coderize=True)
 def bboxes2tblr(priors, gts, normalizer=4.0, normalize_by_wh=True):
     """Encode ground truth boxes to tblr coordinate.
 
@@ -120,7 +118,6 @@ def bboxes2tblr(priors, gts, normalizer=4.0, normalize_by_wh=True):
     return loc / normalizer
 
 
-@mmcv.jit(coderize=True)
 def tblr2bboxes(priors,
                 tblr,
                 normalizer=4.0,