[Fix] Avoid scope switching when using mmdet inference interface (#2039)

Author: Ben-Louis

demo/topdown_demo_with_mmdet.py

@@ -8,13 +8,13 @@
 import mmcv
 import mmengine
 import numpy as np
-from mmengine.registry import init_default_scope
 
 from mmpose.apis import inference_topdown
 from mmpose.apis import init_model as init_pose_estimator
 from mmpose.evaluation.functional import nms
 from mmpose.registry import VISUALIZERS
 from mmpose.structures import merge_data_samples, split_instances
+from mmpose.utils import adapt_mmdet_pipeline
 
 try:
     from mmdet.apis import inference_detector, init_detector
@@ -28,7 +28,6 @@ def process_one_image(args, img_path, detector, pose_estimator, visualizer,
     """Visualize predicted keypoints (and heatmaps) of one image."""
 
     # predict bbox
-    init_default_scope(detector.cfg.get('default_scope', 'mmdet'))
     det_result = inference_detector(detector, img_path)
     pred_instance = det_result.pred_instances.cpu().numpy()
     bboxes = np.concatenate(
@@ -147,6 +146,7 @@ def main():
     # build detector
     detector = init_detector(
         args.det_config, args.det_checkpoint, device=args.device)
+    detector.cfg = adapt_mmdet_pipeline(detector.cfg)
 
     # build pose estimator
     pose_estimator = init_pose_estimator(

mmpose/apis/webcam/nodes/model_nodes/detector_node.py

@@ -2,8 +2,8 @@
 from typing import Dict, List, Optional, Union
 
 import numpy as np
-from mmengine.registry import init_default_scope
 
+from mmpose.utils import adapt_mmdet_pipeline
 from ...utils import get_config_path
 from ..node import Node
 from ..registry import NODES
@@ -92,6 +92,7 @@ def __init__(self,
         # Init model
         self.model = init_detector(
             self.model_config, self.model_checkpoint, device=self.device)
+        self.model.cfg = adapt_mmdet_pipeline(self.model.cfg)
 
         # Register buffers
         self.register_input_buffer(input_buffer, 'input', trigger=True)
@@ -109,7 +110,6 @@ def process(self, input_msgs):
 
         img = input_msg.get_image()
 
-        init_default_scope(self.model.cfg.get('default_scope', 'mmdet'))
         preds = inference_detector(self.model, img)
         objects = self._post_process(preds)
         input_msg.update_objects(objects)

mmpose/utils/__init__.py

@@ -1,11 +1,13 @@
 # Copyright (c) OpenMMLab. All rights reserved.
 from .camera import SimpleCamera, SimpleCameraTorch
 from .collect_env import collect_env
+from .config_utils import adapt_mmdet_pipeline
 from .logger import get_root_logger
 from .setup_env import register_all_modules, setup_multi_processes
 from .timer import StopWatch
 
 __all__ = [
     'get_root_logger', 'collect_env', 'StopWatch', 'setup_multi_processes',
-    'register_all_modules', 'SimpleCamera', 'SimpleCameraTorch'
+    'register_all_modules', 'SimpleCamera', 'SimpleCameraTorch',
+    'adapt_mmdet_pipeline'
 ]

mmpose/utils/config_utils.py

@@ -0,0 +1,26 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from mmpose.utils.typing import ConfigDict
+
+
+def adapt_mmdet_pipeline(cfg: ConfigDict) -> ConfigDict:
+    """Converts pipeline types in MMDetection's test dataloader to use the
+    'mmdet' namespace.
+
+    Args:
+        cfg (ConfigDict): Configuration dictionary for MMDetection.
+
+    Returns:
+        ConfigDict: Configuration dictionary with updated pipeline types.
+    """
+    # use lazy import to avoid hard dependence on mmdet
+    from mmdet.datasets import transforms
+
+    if 'test_dataloader' not in cfg:
+        return cfg
+
+    pipeline = cfg.test_dataloader.dataset.pipeline
+    for trans in pipeline:
+        if trans['type'] in dir(transforms):
+            trans['type'] = 'mmdet.' + trans['type']
+
+    return cfg