fix all mypy issues

src/otx/algo/instance_segmentation/mmdet/models/bbox_heads/bbox_head.py

@@ -171,7 +171,7 @@ def predict_by_feat(
         cls_scores: tuple[Tensor],
         bbox_preds: tuple[Tensor],
         batch_img_metas: list[dict],
-        rcnn_test_cfg: ConfigDict | None = None,
+        rcnn_test_cfg: ConfigDict,
         rescale: bool = False,
     ) -> InstanceList:
         """Transform a batch of output features extracted from the head into bbox results.
@@ -224,8 +224,8 @@ def _predict_by_feat_single(
         cls_score: Tensor,
         bbox_pred: Tensor,
         img_meta: dict,
+        rcnn_test_cfg: ConfigDict,
         rescale: bool = False,
-        rcnn_test_cfg: ConfigDict | None = None,
     ) -> InstanceData:
         """Transform a single image's features extracted from the head into bbox results.
 
@@ -278,13 +278,13 @@ def _predict_by_feat_single(
         if rescale and bboxes.size(0) > 0:
             assert img_meta.get("scale_factor") is not None
             scale_factor = [1 / s for s in img_meta["scale_factor"]]
-            bboxes = scale_boxes(bboxes, scale_factor)
+            bboxes = scale_boxes(bboxes, scale_factor)  # type: ignore
 
         # Get the inside tensor when `bboxes` is a box type
         box_dim = bboxes.size(-1)
         bboxes = bboxes.view(num_rois, -1)
 
-        det_bboxes, det_labels = multiclass_nms(
+        det_bboxes, det_labels = multiclass_nms(  # type: ignore
             bboxes,
             scores,
             rcnn_test_cfg.score_thr,

src/otx/algo/instance_segmentation/mmdet/models/bbox_heads/convfc_bbox_head.py

@@ -37,7 +37,7 @@ def __init__(
         *args,
         **kwargs,
     ) -> None:
-        super().__init__(*args, init_cfg=init_cfg, **kwargs)
+        super().__init__(*args, init_cfg=init_cfg, **kwargs)  # type: ignore
         assert num_shared_convs + num_shared_fcs + num_cls_convs + num_cls_fcs + num_reg_convs + num_reg_fcs > 0
         if num_cls_convs > 0 or num_reg_convs > 0:
             assert num_shared_fcs == 0
@@ -137,12 +137,11 @@ def _add_conv_fc_branch(
             last_layer_dim = self.fc_out_channels
         return branch_convs, branch_fcs, last_layer_dim
 
-    def forward(self, x: tuple[Tensor]) -> tuple:
+    def forward(self, x: Tensor) -> tuple:
         """Forward features from the upstream network.
 
         Args:
-            x (tuple[Tensor]): Features from the upstream network, each is
-                a 4D-tensor.
+            x (Tensor): Features from the upstream network, each is a 4D-tensor.
 
         Returns:
             tuple: A tuple of classification scores and bbox prediction.
@@ -173,7 +172,7 @@ def forward(self, x: tuple[Tensor]) -> tuple:
 @MODELS.register_module()
 class Shared2FCBBoxHead(ConvFCBBoxHead):
     def __init__(self, fc_out_channels: int = 1024, *args, **kwargs) -> None:
-        super().__init__(
+        super().__init__(  # type: ignore
             num_shared_convs=0,
             num_shared_fcs=2,
             num_cls_convs=0,

src/otx/algo/instance_segmentation/mmdet/models/dense_heads/anchor_head.py

@@ -9,10 +9,10 @@
 from torch import Tensor
 
 from otx.algo.instance_segmentation.mmdet.models.utils import (
+    ConfigType,
     InstanceList,
-    OptConfigType,
+    MultiConfig,
     OptInstanceList,
-    OptMultiConfig,
     images_to_levels,
     multi_apply,
     unmap,
@@ -80,16 +80,16 @@ class AnchorHead(BaseDenseHead):
     def __init__(
         self,
         num_classes: int,
-        in_channels: tuple[int, ...],
+        in_channels: int,
         anchor_generator: dict,
         bbox_coder: dict,
         loss_cls: dict,
         loss_bbox: dict,
+        train_cfg: ConfigType,
+        test_cfg: ConfigType,
+        init_cfg: MultiConfig,
         feat_channels: int = 256,
         reg_decoded_bbox: bool = False,
-        train_cfg: OptConfigType = None,
-        test_cfg: OptConfigType = None,
-        init_cfg: OptMultiConfig = None,
     ) -> None:
         super().__init__(init_cfg=init_cfg)
         self.in_channels = in_channels
@@ -142,7 +142,7 @@ def forward(self, x: tuple[Tensor]) -> tuple[list[Tensor]]:
                     scale levels, each is a 4D-tensor, the channels number \
                     is num_base_priors * 4.
         """
-        return multi_apply(self.forward_single, x)
+        return multi_apply(self.forward_single, x)  # type: ignore
 
     def get_anchors(
         self,
@@ -389,10 +389,6 @@ def get_targets(
         bbox_targets_list = images_to_levels(all_bbox_targets, num_level_anchors)
         bbox_weights_list = images_to_levels(all_bbox_weights, num_level_anchors)
         res = (labels_list, label_weights_list, bbox_targets_list, bbox_weights_list, avg_factor)
-        if return_sampling_results:
-            res = (*res, sampling_results_list)
-        for i, r in enumerate(rest_results):  # user-added return values
-            rest_results[i] = images_to_levels(r, num_level_anchors)
 
         return res + tuple(rest_results)
 

src/otx/algo/instance_segmentation/mmdet/models/dense_heads/base_dense_head.py

@@ -3,7 +3,7 @@
 # Copyright (c) OpenMMLab. All rights reserved.
 from __future__ import annotations
 
-from abc import ABCMeta, abstractmethod
+from abc import ABCMeta
 from typing import TYPE_CHECKING
 
 # TODO(Eugene): replace mmcv.batched_nms with torchvision
@@ -77,10 +77,6 @@ def init_weights(self) -> None:
             if hasattr(m, "conv_offset"):
                 constant_init(m.conv_offset, 0)
 
-    @abstractmethod
-    def loss_by_feat(self, **kwargs) -> dict:
-        """Calculate the loss based on the features extracted by the detection head."""
-
     def loss_and_predict(
         self,
         x: tuple[Tensor],
@@ -142,8 +138,8 @@ def predict_by_feat(
         self,
         cls_scores: list[Tensor],
         bbox_preds: list[Tensor],
+        batch_img_metas: list[dict],
         score_factors: list[Tensor] | None = None,
-        batch_img_metas: list[dict] | None = None,
         cfg: ConfigDict | None = None,
         rescale: bool = False,
         with_nms: bool = True,

src/otx/algo/instance_segmentation/mmdet/models/dense_heads/rpn_head.py

@@ -23,7 +23,6 @@
 from otx.algo.instance_segmentation.mmdet.structures.bbox import (
     empty_box_as,
     get_box_wh,
-    scale_boxes,
 )
 
 from .anchor_head import AnchorHead
@@ -266,12 +265,8 @@ def _bbox_post_process(
             raise RuntimeError(msg)
 
         if rescale:
-            if img_meta.get("scale_factor") is None:
-                msg = "scale_factor is required when rescale is True"
-                raise ValueError(msg)
-
-            scale_factor = [1 / s for s in img_meta["scale_factor"]]
-            results.bboxes = scale_boxes(results.bboxes, scale_factor)
+            msg = "Rescale is not implemented in RPNHead"
+            raise NotImplementedError
 
         # filter small size bboxes
         if cfg.get("min_bbox_size", -1) >= 0:

src/otx/algo/instance_segmentation/mmdet/models/detectors/base.py

@@ -95,7 +95,7 @@ def predict(self, batch_inputs: Tensor, batch_data_samples: SampleList) -> Sampl
         """Predict results from a batch of inputs and data samples with post-processing."""
 
     @abstractmethod
-    def _forward(self, batch_inputs: Tensor, batch_data_samples: OptSampleList = None) -> tuple:
+    def _forward(self, batch_inputs: Tensor, batch_data_samples: SampleList) -> tuple:
         """Network forward process.
 
         Usually includes backbone, neck and head forward without any post-

src/otx/algo/instance_segmentation/mmdet/models/detectors/two_stage.py

@@ -30,57 +30,54 @@ class TwoStageDetector(BaseDetector):
     def __init__(
         self,
         backbone: ConfigType,
-        neck: OptConfigType = None,
-        rpn_head: OptConfigType = None,
-        roi_head: OptConfigType = None,
-        train_cfg: OptConfigType = None,
-        test_cfg: OptConfigType = None,
+        neck: ConfigType,
+        rpn_head: ConfigType,
+        roi_head: ConfigType,
+        train_cfg: ConfigType,
+        test_cfg: ConfigType,
         data_preprocessor: OptConfigType = None,
         init_cfg: OptMultiConfig = None,
     ) -> None:
         super().__init__(data_preprocessor=data_preprocessor, init_cfg=init_cfg)
         self.backbone = MODELS.build(backbone)
 
-        if neck is not None:
-            if neck.type != FPN.__name__:
-                msg = f"neck type must be {FPN.__name__}, but got {neck.type}"
-                raise ValueError(msg)
-            # pop out type for FPN
-            neck.pop("type")
-            self.neck = FPN(**neck)
-
-        if rpn_head is not None:
-            rpn_train_cfg = train_cfg.rpn if train_cfg is not None else None
-            rpn_head_ = rpn_head.copy()
-            rpn_head_.update(train_cfg=rpn_train_cfg, test_cfg=test_cfg.rpn)
-            rpn_head_num_classes = rpn_head_.get("num_classes", None)
-            if rpn_head_num_classes is None:
-                rpn_head_.update(num_classes=1)
-            elif rpn_head_num_classes != 1:
-                warnings.warn(
-                    "The `num_classes` should be 1 in RPN, but get "
-                    f"{rpn_head_num_classes}, please set "
-                    "rpn_head.num_classes = 1 in your config file.",
-                )
-                rpn_head_.update(num_classes=1)
-            if rpn_head_.type != RPNHead.__name__:
-                msg = f"rpn_head type must be {RPNHead.__name__}, but got {rpn_head_.type}"
-                raise ValueError(msg)
-            # pop out type for RPNHead
-            rpn_head_.pop("type")
-            self.rpn_head = RPNHead(**rpn_head_)
-
-        if roi_head is not None:
-            # update train and test cfg here for now
-            rcnn_train_cfg = train_cfg.rcnn if train_cfg is not None else None
-            roi_head.update(train_cfg=rcnn_train_cfg)
-            roi_head.update(test_cfg=test_cfg.rcnn)
-            if roi_head.type != CustomRoIHead.__name__:
-                msg = f"roi_head type must be {CustomRoIHead.__name__}, but got {roi_head.type}"
-                raise ValueError(msg)
-            # pop out type for RoIHead
-            roi_head.pop("type")
-            self.roi_head = CustomRoIHead(**roi_head)
+        if neck["type"] != FPN.__name__:
+            msg = f"neck type must be {FPN.__name__}, but got {neck['type']}"
+            raise ValueError(msg)
+        # pop out type for FPN
+        neck.pop("type")
+        self.neck = FPN(**neck)
+
+        rpn_train_cfg = train_cfg["rpn"]
+        rpn_head_ = rpn_head.copy()
+        rpn_head_.update(train_cfg=rpn_train_cfg, test_cfg=test_cfg["rpn"])
+        rpn_head_num_classes = rpn_head_.get("num_classes", None)
+        if rpn_head_num_classes is None:
+            rpn_head_.update(num_classes=1)
+        elif rpn_head_num_classes != 1:
+            warnings.warn(
+                "The `num_classes` should be 1 in RPN, but get "
+                f"{rpn_head_num_classes}, please set "
+                "rpn_head.num_classes = 1 in your config file.",
+            )
+            rpn_head_.update(num_classes=1)
+        if rpn_head_["type"] != RPNHead.__name__:
+            msg = f"rpn_head type must be {RPNHead.__name__}, but got {rpn_head_['type']}"
+            raise ValueError(msg)
+        # pop out type for RPNHead
+        rpn_head_.pop("type")
+        self.rpn_head = RPNHead(**rpn_head_)
+
+        # update train and test cfg here for now
+        rcnn_train_cfg = train_cfg["rcnn"]
+        roi_head.update(train_cfg=rcnn_train_cfg)
+        roi_head.update(test_cfg=test_cfg["rcnn"])
+        if roi_head["type"] != CustomRoIHead.__name__:
+            msg = f"roi_head type must be {CustomRoIHead.__name__}, but got {roi_head['type']}"
+            raise ValueError(msg)
+        # pop out type for RoIHead
+        roi_head.pop("type")
+        self.roi_head = CustomRoIHead(**roi_head)
 
         self.train_cfg = train_cfg
         self.test_cfg = test_cfg
@@ -158,7 +155,7 @@ def _forward(self, batch_inputs: Tensor, batch_data_samples: SampleList) -> tupl
                 raise ValueError(msg)
             rpn_results_list = [data_sample.proposals for data_sample in batch_data_samples]
         roi_outs = self.roi_head.forward(x, rpn_results_list, batch_data_samples)
-        return results + (roi_outs,)
+        return (*results, roi_outs)
 
     def loss(self, batch_inputs: Tensor, batch_data_samples: SampleList) -> dict:
         """Calculate losses from a batch of inputs and data samples.
@@ -175,11 +172,11 @@ def loss(self, batch_inputs: Tensor, batch_data_samples: SampleList) -> dict:
         """
         x = self.extract_feat(batch_inputs)
 
-        losses = dict()
+        losses = {}
 
         # RPN forward and loss
         if self.with_rpn:
-            proposal_cfg = self.train_cfg.get("rpn_proposal", self.test_cfg.rpn)
+            proposal_cfg = self.train_cfg.get("rpn_proposal", self.test_cfg["rpn"])
             rpn_data_samples = copy.deepcopy(batch_data_samples)
             # set cat_id of gt_labels to 0 in RPN
             for data_sample in rpn_data_samples:

src/otx/algo/instance_segmentation/mmdet/models/layers/bbox_nms.py

@@ -1,7 +1,8 @@
 """The original source code is from mmdet. Please refer to https://github.com/open-mmlab/mmdetection/."""
 
 # Copyright (c) OpenMMLab. All rights reserved.
-from typing import Optional, Tuple, Union
+
+from __future__ import annotations
 
 import torch
 
@@ -19,10 +20,10 @@ def multiclass_nms(
     score_thr: float,
     nms_cfg: ConfigType,
     max_num: int = -1,
-    score_factors: Optional[Tensor] = None,
+    score_factors: Tensor | None = None,
     return_inds: bool = False,
     box_dim: int = 4,
-) -> Union[Tuple[Tensor, Tensor, Tensor], Tuple[Tensor, Tensor]]:
+) -> tuple[Tensor, Tensor, Tensor] | tuple[Tensor, Tensor]:
     """NMS for multi-class bboxes.
 
     Args:
@@ -91,8 +92,7 @@ def multiclass_nms(
         dets = torch.cat([bboxes, scores[:, None]], -1)
         if return_inds:
             return dets, labels, inds
-        else:
-            return dets, labels
+        return dets, labels
 
     dets, keep = batched_nms(bboxes, scores, labels, nms_cfg)
 
@@ -102,5 +102,4 @@ def multiclass_nms(
 
     if return_inds:
         return dets, labels[keep], inds[keep]
-    else:
-        return dets, labels[keep]
+    return dets, labels[keep]