mmclassification ConformerHead support (#1905)

* mmclassification ConformerHead support

* add mmclassification ConformerHead test config

---------

Co-authored-by: lishengxi <[email protected]>

Author: xizi

mmdeploy/codebase/mmcls/models/classifiers/base.py

@@ -35,9 +35,11 @@ def base_classifier__forward(
     if self.head is not None:
         output = self.head(output)
 
-    from mmcls.models.heads import MultiLabelClsHead
+    from mmcls.models.heads import ConformerHead, MultiLabelClsHead
     if isinstance(self.head, MultiLabelClsHead):
         output = torch.sigmoid(output)
+    elif isinstance(self.head, ConformerHead):
+        output = F.softmax(torch.add(output[0], output[1]), dim=1)
     else:
         output = F.softmax(output, dim=1)
     return output

tests/regression/mmcls.yml

@@ -228,3 +228,11 @@ models:
       - *pipeline_ort_static_fp32
       - convert_image: *convert_image
         deploy_config: configs/mmcls/classification_tensorrt_dynamic-224x224-224x224.py
+
+  - name: Conformer
+    metafile: configs/conformer/metafile.yml
+    model_configs:
+      - configs/conformer/conformer-tiny-p16_8xb128_in1k.py
+    pipelines:
+      - *pipeline_ort_dynamic_fp32
+      - *pipeline_trt_dynamic_fp32