[FEATURE] Add MoViNet template for action classification (#1742)

* Add MoViNet template for action classification
Co-authored-by: Nikita Savelyev <nikita.savelyev@intel.com>

Author: JihwanEom

otx/algorithms/action/adapters/mmaction/models/__init__.py

@@ -3,8 +3,9 @@
 # Copyright (C) 2022 Intel Corporation
 # SPDX-License-Identifier: Apache-2.0
 
-from .backbones import register_action_backbones
+from .backbones import OTXMoViNet, register_action_backbones
 from .detectors import AVAFastRCNN
-from .heads import AVARoIHead
+from .heads import AVARoIHead, MoViNetHead
+from .recognizers import MoViNetRecognizer
 
-__all__ = ["register_action_backbones", "AVAFastRCNN", "AVARoIHead"]
+__all__ = ["register_action_backbones", "AVAFastRCNN", "OTXMoViNet", "MoViNetHead", "MoViNetRecognizer", "AVARoIHead"]

otx/algorithms/action/adapters/mmaction/models/backbones/__init__.py

@@ -6,7 +6,12 @@
 from mmaction.models.backbones.x3d import X3D
 from mmdet.models import BACKBONES as MMDET_BACKBONES
 
+from .movinet import OTXMoViNet
+
 
 def register_action_backbones():
     """Register action backbone to mmdetection backbones."""
     MMDET_BACKBONES.register_module()(X3D)
+
+
+__all__ = ["OTXMoViNet"]