refactor dump-info (#1040)

mmdeploy/backend/sdk/export_info.py

@@ -6,9 +6,9 @@ from typing import Dict, List, Tuple, Union
 import mmengine
 
 from mmdeploy.apis import build_task_processor
-from mmdeploy.utils import (Backend, Task, get_backend, get_codebase,
-                            get_common_config, get_ir_config, get_root_logger,
-                            get_task_type, is_dynamic_batch, load_config)
+from mmdeploy.utils import (Task, get_backend, get_codebase, get_ir_config,
+                            get_precision, get_root_logger, get_task_type,
+                            is_dynamic_batch, load_config)
 from mmdeploy.utils.constants import SDK_TASK_MAP as task_map
 
 
@@ -79,60 +79,31 @@ def get_models(deploy_cfg: Union[str, mmengine.Config],
     name, _ = get_model_name_customs(deploy_cfg, model_cfg, work_dir)
     precision = 'FP32'
     ir_name = get_ir_config(deploy_cfg)['save_file']
-    net = ir_name
     weights = ''
-    backend = get_backend(deploy_cfg=deploy_cfg)
+    backend = get_backend(deploy_cfg=deploy_cfg).value
 
-    def replace_suffix(file_name: str, dst_suffix: str) -> str:
-        """Replace the suffix to the destination one.
-
-        Args:
-            file_name (str): The file name to be operated.
-            dst_suffix (str): The destination suffix.
-
-        Return:
-            str: The file name of which the suffix has been replaced.
-        """
-        return re.sub(r'\.[a-z]+', dst_suffix, file_name)
-
-    if backend == Backend.TENSORRT:
-        net = replace_suffix(ir_name, '.engine')
-        common_cfg = get_common_config(deploy_cfg)
-        fp16_mode = common_cfg.get('fp16_mode', False)
-        int8_mode = common_cfg.get('int8_mode', False)
-        if fp16_mode:
-            precision = 'FP16'
-        if int8_mode:
-            precision = 'INT8'
-    elif backend == Backend.PPLNN:
-        precision = 'FP16'
-        weights = replace_suffix(ir_name, '.json')
-        net = ir_name
-    elif backend == Backend.OPENVINO:
-        net = replace_suffix(ir_name, '.xml')
-        weights = replace_suffix(ir_name, '.bin')
-    elif backend == Backend.NCNN:
-        net = replace_suffix(ir_name, '.param')
-        weights = replace_suffix(ir_name, '.bin')
-        if 'precision' in deploy_cfg['backend_config']:
-            precision = deploy_cfg['backend_config']['precision']
-    elif backend == Backend.SNPE:
-        net = replace_suffix(ir_name, '.dlc')
-    elif backend in [Backend.ONNXRUNTIME, Backend.TORCHSCRIPT]:
-        pass
-    else:
-        raise NotImplementedError(f'Not supported backend: {backend.value}.')
+    backend_net = dict(
+        tensorrt=lambda file: re.sub(r'\.[a-z]+', '.engine', file),
+        openvino=lambda file: re.sub(r'\.[a-z]+', '.xml', file),
+        ncnn=lambda file: re.sub(r'\.[a-z]+', '.param', file),
+        snpe=lambda file: re.sub(r'\.[a-z]+', '.dlc', file))
+    backend_weights = dict(
+        pplnn=lambda file: re.sub(r'\.[a-z]+', '.json', file),
+        openvino=lambda file: re.sub(r'\.[a-z]+', '.bin', file),
+        ncnn=lambda file: re.sub(r'\.[a-z]+', '.bin', file))
+    net = backend_net.get(backend, lambda x: x)(ir_name)
+    weights = backend_weights.get(backend, lambda x: weights)(ir_name)
 
+    precision = get_precision(deploy_cfg)
     dynamic_shape = is_dynamic_batch(deploy_cfg, input_name='input')
-    batch_size = 1
     return [
         dict(
             name=name,
             net=net,
             weights=weights,
-            backend=backend.value,
+            backend=backend,
             precision=precision,
-            batch_size=batch_size,
+            batch_size=1,
             dynamic_shape=dynamic_shape)
     ]
 
@@ -151,20 +122,16 @@ def get_inference_info(deploy_cfg: mmengine.Config, model_cfg: mmengine.Config,
             input_map.
     """
     name, _ = get_model_name_customs(deploy_cfg, model_cfg, work_dir)
-    type = 'Task'
-    module = 'Net'
-    input = ['prep_output']
-    output = ['infer_output']
     ir_config = get_ir_config(deploy_cfg)
     input_names = ir_config.get('input_names', None)
     input_name = input_names[0] if input_names else 'input'
     input_map = dict(img=input_name)
     return_dict = dict(
         name=name,
-        type=type,
-        module=module,
-        input=input,
-        output=output,
+        type='Task',
+        module='Net',
+        input=['prep_output'],
+        output=['infer_output'],
         input_map=input_map)
     if 'use_vulkan' in deploy_cfg['backend_config']:
         return_dict['use_vulkan'] = deploy_cfg['backend_config']['use_vulkan']
@@ -174,71 +141,15 @@ def get_inference_info(deploy_cfg: mmengine.Config, model_cfg: mmengine.Config,
 def get_preprocess(deploy_cfg: mmengine.Config, model_cfg: mmengine.Config):
     task_processor = build_task_processor(
         model_cfg=model_cfg, deploy_cfg=deploy_cfg, device='cpu')
-    pipeline = task_processor.get_preprocess()
-    type = 'Task'
-    module = 'Transform'
-    name = 'Preprocess'
-    input = ['img']
-    output = ['prep_output']
-    meta_keys = [
-        'filename', 'ori_filename', 'ori_shape', 'img_shape', 'pad_shape',
-        'scale_factor', 'flip', 'flip_direction', 'img_norm_cfg', 'valid_ratio'
-    ]
-    transforms = [
-        item for item in pipeline
-        if 'Random' not in item['type'] and 'RescaleToZeroOne' not in
-        item['type'] and 'Annotation' not in item['type']
-    ]
-    for i, transform in enumerate(transforms):
-        if 'keys' in transform and transform['keys'] == ['lq']:
-            transform['keys'] = ['img']
-        if 'key' in transform and transform['key'] == 'lq':
-            transform['key'] = 'img'
-        if transform['type'] == 'ResizeEdge':
-            transform['type'] = 'Resize'
-            transform['keep_ratio'] = True
-            # now the sdk of class has bugs, because ResizeEdge not implement
-            # in sdk.
-            transform['size'] = (transform['scale'], transform['scale'])
-        if transform['type'] in ('PackTextDetInputs', 'PackTextRecogInputs'):
-            meta_keys += transform[
-                'meta_keys'] if 'meta_keys' in transform else []
-            transform['meta_keys'] = list(set(meta_keys))
-            transform['keys'] = ['img']
-            transforms[i]['type'] = 'Collect'
-        if transform['type'] == 'PackDetInputs' or \
-           transform['type'] == 'PackClsInputs':
-            transforms.insert(i, dict(type='DefaultFormatBundle'))
-            transform['type'] = 'Collect'
-            if 'keys' not in transform:
-                transform['keys'] = ['img']
-        if transform['type'] == 'Resize':
-            transforms[i]['size'] = transforms[i]['scale']
-
-    data_preprocessor = model_cfg.model.data_preprocessor
-    transforms.insert(-1, dict(type='DefaultFormatBundle'))
-    transforms.insert(
-        -2,
-        dict(
-            type='Pad',
-            size_divisor=data_preprocessor.get('pad_size_divisor', 1)))
-    transforms.insert(
-        -3,
-        dict(
-            type='Normalize',
-            to_rgb=data_preprocessor.get('bgr_to_rgb', False),
-            mean=data_preprocessor.get('mean', [0, 0, 0]),
-            std=data_preprocessor.get('std', [1, 1, 1])))
-
-    assert transforms[0]['type'] == 'LoadImageFromFile', 'The first item type'\
-        ' of pipeline should be LoadImageFromFile'
-
+    transforms = task_processor.get_preprocess()
+    assert transforms[0]['type'] == 'LoadImageFromFile', 'The first item'\
+        ' type of pipeline should be LoadImageFromFile'
     return dict(
-        type=type,
-        module=module,
-        name=name,
-        input=input,
-        output=output,
+        type='Task',
+        module='Transform',
+        name='Preprocess',
+        input=['img'],
+        output=['prep_output'],
         transforms=transforms)
 
 
@@ -255,38 +166,17 @@ def get_postprocess(deploy_cfg: mmengine.Config, model_cfg: mmengine.Config,
         dict: Composed of the model name, type, module, input, params and
             output.
     """
-    module = get_codebase(deploy_cfg).value
-    type = 'Task'
-    name = 'postprocess'
-    params = dict()
-    task = get_task_type(deploy_cfg)
     task_processor = build_task_processor(
         model_cfg=model_cfg, deploy_cfg=deploy_cfg, device='cpu')
-    post_processor = task_processor.get_postprocess()
+    post_processor = task_processor.get_postprocess(work_dir)
 
-    # TODO remove after adding instance segmentation to task processor
-    if task == Task.OBJECT_DETECTION and 'mask_thr_binary' in post_processor:
-        task = Task.INSTANCE_SEGMENTATION
-
-    component = task_map[task]['component']
-    if task not in (Task.SUPER_RESOLUTION, Task.SEGMENTATION):
-        if 'type' in post_processor:
-            component = post_processor.pop('type')
-    output = ['post_output']
-
-    if task == Task.TEXT_RECOGNITION:
-        import shutil
-        shutil.copy(model_cfg.dictionary.dict_file,
-                    f'{work_dir}/dict_file.txt')
-        with_padding = model_cfg.dictionary.get('with_padding', False)
-        params = dict(dict_file='dict_file.txt', with_padding=with_padding)
     return dict(
-        type=type,
-        module=module,
-        name=name,
-        component=component,
-        params=params,
-        output=output)
+        type='Task',
+        module=get_codebase(deploy_cfg).value,
+        name='postprocess',
+        component=post_processor['type'],
+        params=post_processor.get('params', dict()),
+        output=['post_output'])
 
 
 def get_deploy(deploy_cfg: mmengine.Config, model_cfg: mmengine.Config,
@@ -384,7 +274,6 @@ def export2SDK(deploy_cfg: Union[str, mmengine.Config],
         pth (str): The path of the model checkpoint weights.
     """
     deploy_cfg, model_cfg = load_config(deploy_cfg, model_cfg)
-    print(f'debugging what is type(model_cfg): {type(model_cfg)}')
     deploy_info = get_deploy(deploy_cfg, model_cfg, work_dir=work_dir)
     pipeline_info = get_pipeline(deploy_cfg, model_cfg, work_dir=work_dir)
     detail_info = get_detail(deploy_cfg, model_cfg, pth=pth)

mmdeploy/codebase/base/task.py

@@ -322,7 +322,7 @@ class BaseTask(metaclass=ABCMeta):
         return input_data['inputs']
 
     @abstractmethod
-    def get_preprocess(self) -> Dict:
+    def get_preprocess(self, *args, **kwargs) -> Dict:
         """Get the preprocess information for SDK.
 
         Return:
@@ -331,7 +331,7 @@ class BaseTask(metaclass=ABCMeta):
         pass
 
     @abstractmethod
-    def get_postprocess(self) -> Dict:
+    def get_postprocess(self, *args, **kwargs) -> Dict:
         """Get the postprocess information for SDK.
 
         Return:
@@ -340,7 +340,7 @@ class BaseTask(metaclass=ABCMeta):
         pass
 
     @abstractmethod
-    def get_model_name(self) -> str:
+    def get_model_name(self, *args, **kwargs) -> str:
         """Get the model name.
 
         Return:

mmdeploy/codebase/mmcls/deploy/classification.py

@@ -241,7 +241,7 @@ class Classification(BaseTask):
         """
         raise NotImplementedError('Not supported yet.')
 
-    def get_preprocess(self) -> Dict:
+    def get_preprocess(self, *args, **kwargs) -> Dict:
         """Get the preprocess information for SDK.
 
         Return:
@@ -252,7 +252,7 @@ class Classification(BaseTask):
         preprocess = cfg.test_pipeline
         return preprocess
 
-    def get_postprocess(self) -> Dict:
+    def get_postprocess(self, *args, **kwargs) -> Dict:
         """Get the postprocess information for SDK.
 
         Return:
@@ -269,7 +269,7 @@ class Classification(BaseTask):
         postprocess.topk = max(topk)
         return postprocess
 
-    def get_model_name(self) -> str:
+    def get_model_name(self, *args, **kwargs) -> str:
         """Get the model name.
 
         Return:

mmdeploy/codebase/mmdet/deploy/object_detection.py

@@ -206,7 +206,7 @@ class ObjectDetection(BaseTask):
             f'Unknown partition_type {partition_type}'
         return MMDET_PARTITION_CFG[partition_type]
 
-    def get_preprocess(self) -> Dict:
+    def get_preprocess(self, *args, **kwargs) -> Dict:
         """Get the preprocess information for SDK.
 
         Return:
@@ -217,7 +217,7 @@ class ObjectDetection(BaseTask):
         preprocess = model_cfg.test_pipeline
         return preprocess
 
-    def get_postprocess(self) -> Dict:
+    def get_postprocess(self, *args, **kwargs) -> Dict:
         """Get the postprocess information for SDK.
 
         Return:
@@ -233,7 +233,7 @@ class ObjectDetection(BaseTask):
                     'mask_thr_binary']
         return postprocess
 
-    def get_model_name(self) -> str:
+    def get_model_name(self, *args, **kwargs) -> str:
         """Get the model name.
 
         Return:

mmdeploy/codebase/mmdet3d/deploy/voxel_detection.py

@@ -199,7 +199,7 @@ class VoxelDetection(BaseTask):
             eval_kwargs.update(dict(metric=metrics, **kwargs))
             dataset.evaluate(outputs, **eval_kwargs)
 
-    def get_model_name(self) -> str:
+    def get_model_name(self, *args, **kwargs) -> str:
         """Get the model name.
 
         Return:
@@ -230,7 +230,7 @@ class VoxelDetection(BaseTask):
         """
         raise NotImplementedError
 
-    def get_postprocess(self) -> Dict:
+    def get_postprocess(self, *args, **kwargs) -> Dict:
         """Get the postprocess information for SDK.
 
         Return:
@@ -238,7 +238,7 @@ class VoxelDetection(BaseTask):
         """
         raise NotImplementedError
 
-    def get_preprocess(self) -> Dict:
+    def get_preprocess(self, *args, **kwargs) -> Dict:
         """Get the preprocess information for SDK.
 
         Return:

mmdeploy/codebase/mmedit/deploy/super_resolution.py

@@ -286,7 +286,7 @@ class SuperResolution(BaseTask):
         for stat in stats:
             logger.info('Eval-{}: {}'.format(stat, stats[stat]))
 
-    def get_preprocess(self) -> Dict:
+    def get_preprocess(self, *args, **kwargs) -> Dict:
         """Get the preprocess information for SDK.
 
         Return:
@@ -300,7 +300,7 @@ class SuperResolution(BaseTask):
                 item['std'] = [255, 255, 255]
         return preprocess
 
-    def get_postprocess(self) -> Dict:
+    def get_postprocess(self, *args, **kwargs) -> Dict:
         """Get the postprocess information for SDK.
 
         Return:
@@ -308,7 +308,7 @@ class SuperResolution(BaseTask):
         """
         return dict()
 
-    def get_model_name(self) -> str:
+    def get_model_name(self, *args, **kwargs) -> str:
         """Get the model name.
 
         Return:

mmdeploy/codebase/mmocr/deploy/text_detection.py

@@ -212,7 +212,7 @@ class TextDetection(BaseTask):
         """
         raise NotImplementedError('Not supported yet.')
 
-    def get_preprocess(self) -> Dict:
+    def get_preprocess(self, *args, **kwargs) -> Dict:
         """Get the preprocess information for SDK.
 
         Return:
@@ -220,10 +220,43 @@ class TextDetection(BaseTask):
         """
         input_shape = get_input_shape(self.deploy_cfg)
         model_cfg = process_model_config(self.model_cfg, [''], input_shape)
-        preprocess = model_cfg.test_dataloader.dataset.pipeline
-        return preprocess
+        pipeline = model_cfg.test_dataloader.dataset.pipeline
+        meta_keys = [
+            'filename', 'ori_filename', 'ori_shape', 'img_shape', 'pad_shape',
+            'scale_factor', 'flip', 'flip_direction', 'img_norm_cfg',
+            'valid_ratio'
+        ]
+        transforms = [
+            item for item in pipeline if 'Random' not in item['type']
+            and 'Annotation' not in item['type']
+        ]
+        for i, transform in enumerate(transforms):
+            if transform['type'] == 'PackTextDetInputs':
+                meta_keys += transform[
+                    'meta_keys'] if 'meta_keys' in transform else []
+                transform['meta_keys'] = list(set(meta_keys))
+                transform['keys'] = ['img']
+                transforms[i]['type'] = 'Collect'
+            if transform['type'] == 'Resize':
+                transforms[i]['size'] = transforms[i]['scale']
 
-    def get_postprocess(self) -> Dict:
+        data_preprocessor = model_cfg.model.data_preprocessor
+        transforms.insert(-1, dict(type='DefaultFormatBundle'))
+        transforms.insert(
+            -2,
+            dict(
+                type='Pad',
+                size_divisor=data_preprocessor.get('pad_size_divisor', 1)))
+        transforms.insert(
+            -3,
+            dict(
+                type='Normalize',
+                to_rgb=data_preprocessor.get('bgr_to_rgb', False),
+                mean=data_preprocessor.get('mean', [0, 0, 0]),
+                std=data_preprocessor.get('std', [1, 1, 1])))
+        return transforms
+
+    def get_postprocess(self, *args, **kwargs) -> Dict:
         """Get the postprocess information for SDK.
 
         Return:
@@ -232,7 +265,7 @@ class TextDetection(BaseTask):
         postprocess = self.model_cfg.model.det_head
         return postprocess
 
-    def get_model_name(self) -> str:
+    def get_model_name(self, *args, **kwargs) -> str:
         """Get the model name.
 
         Return:

mmdeploy/codebase/mmocr/deploy/text_recognition.py

@@ -218,7 +218,7 @@ class TextRecognition(BaseTask):
         """
         raise NotImplementedError('Not supported yet.')
 
-    def get_preprocess(self) -> Dict:
+    def get_preprocess(self, *args, **kwargs) -> Dict:
         """Get the preprocess information for SDK.
 
         Return:
@@ -226,21 +226,62 @@ class TextRecognition(BaseTask):
         """
         input_shape = get_input_shape(self.deploy_cfg)
         model_cfg = process_model_config(self.model_cfg, [''], input_shape)
-        preprocess = model_cfg.test_dataloader.dataset.pipeline
-        return preprocess
+        pipeline = model_cfg.test_dataloader.dataset.pipeline
+        meta_keys = [
+            'filename', 'ori_filename', 'ori_shape', 'img_shape', 'pad_shape',
+            'scale_factor', 'flip', 'flip_direction', 'img_norm_cfg',
+            'valid_ratio'
+        ]
+        transforms = [
+            item for item in pipeline if 'Random' not in item['type']
+            and 'Annotation' not in item['type']
+        ]
+        for i, transform in enumerate(transforms):
+            if transform['type'] == 'PackTextRecogInputs':
+                meta_keys += transform[
+                    'meta_keys'] if 'meta_keys' in transform else []
+                transform['meta_keys'] = list(set(meta_keys))
+                transform['keys'] = ['img']
+                transforms[i]['type'] = 'Collect'
+            if transform['type'] == 'Resize':
+                transforms[i]['size'] = transforms[i]['scale']
 
-    def get_postprocess(self) -> Dict:
+        data_preprocessor = model_cfg.model.data_preprocessor
+        transforms.insert(-1, dict(type='DefaultFormatBundle'))
+        transforms.insert(
+            -2,
+            dict(
+                type='Pad',
+                size_divisor=data_preprocessor.get('pad_size_divisor', 1)))
+        transforms.insert(
+            -3,
+            dict(
+                type='Normalize',
+                to_rgb=data_preprocessor.get('bgr_to_rgb', False),
+                mean=data_preprocessor.get('mean', [0, 0, 0]),
+                std=data_preprocessor.get('std', [1, 1, 1])))
+        return transforms
+
+    def get_postprocess(self,
+                        work_dir: Optional[str] = None,
+                        **kwargs) -> Dict:
         """Get the postprocess information for SDK.
 
         Return:
-            dict: Composed of the postprocess information.
+            Dict: Composed of the postprocess information.
         """
         postprocess = self.model_cfg.model.decoder.postprocessor
         if postprocess.type == 'CTCPostProcessor':
             postprocess.type = 'CTCConvertor'
+        import shutil
+        shutil.copy(self.model_cfg.dictionary.dict_file,
+                    f'{work_dir}/dict_file.txt')
+        with_padding = self.model_cfg.dictionary.get('with_padding', False)
+        params = dict(dict_file='dict_file.txt', with_padding=with_padding)
+        postprocess['params'] = params
         return postprocess
 
-    def get_model_name(self) -> str:
+    def get_model_name(self, *args, **kwargs) -> str:
         """Get the model name.
 
         Return:

mmdeploy/codebase/mmpose/deploy/pose_detection.py

@@ -305,7 +305,7 @@ class PoseDetection(BaseTask):
         for k, v in sorted(results.items()):
             logger.info(f'{k}: {v:.4f}')
 
-    def get_model_name(self) -> str:
+    def get_model_name(self, *args, **kwargs) -> str:
         """Get the model name.
 
         Return:
@@ -324,7 +324,7 @@ class PoseDetection(BaseTask):
         """
         raise NotImplementedError('Not supported yet.')
 
-    def get_preprocess(self) -> Dict:
+    def get_preprocess(self, *args, **kwargs) -> Dict:
         """Get the preprocess information for SDK.
 
         Return:
@@ -335,7 +335,7 @@ class PoseDetection(BaseTask):
         preprocess = model_cfg.data.test.pipeline
         return preprocess
 
-    def get_postprocess(self) -> Dict:
+    def get_postprocess(self, *args, **kwargs) -> Dict:
         """Get the postprocess information for SDK."""
         postprocess = {'type': 'UNKNOWN'}
         if self.model_cfg.model.type == 'TopDown':

mmdeploy/codebase/mmrotate/deploy/rotated_detection.py

@@ -335,7 +335,7 @@ class RotatedDetection(BaseTask):
             eval_kwargs.update(dict(metric=metrics, **kwargs))
             logger.info(dataset.evaluate(outputs, **eval_kwargs))
 
-    def get_preprocess(self) -> Dict:
+    def get_preprocess(self, *args, **kwargs) -> Dict:
         """Get the preprocess information for SDK.
 
         Return:
@@ -349,7 +349,7 @@ class RotatedDetection(BaseTask):
         preprocess = model_cfg.data.test.pipeline
         return preprocess
 
-    def get_postprocess(self) -> Dict:
+    def get_postprocess(self, *args, **kwargs) -> Dict:
         """Get the postprocess information for SDK.
 
         Return:
@@ -358,7 +358,7 @@ class RotatedDetection(BaseTask):
         postprocess = self.model_cfg.model.test_cfg
         return postprocess
 
-    def get_model_name(self) -> str:
+    def get_model_name(self, *args, **kwargs) -> str:
         """Get the model name.
 
         Return:

mmdeploy/codebase/mmseg/deploy/segmentation.py

@@ -241,7 +241,7 @@ class Segmentation(BaseTask):
     def get_partition_cfg(partition_type: str) -> Dict:
         raise NotImplementedError('Not supported yet.')
 
-    def get_preprocess(self) -> Dict:
+    def get_preprocess(self, *args, **kwargs) -> Dict:
         """Get the preprocess information for SDK.
 
         Return:
@@ -272,7 +272,7 @@ class Segmentation(BaseTask):
                 ]))
         return preprocess
 
-    def get_postprocess(self) -> Dict:
+    def get_postprocess(self, *args, **kwargs) -> Dict:
         """Get the postprocess information for SDK.
 
         Return:
@@ -281,7 +281,7 @@ class Segmentation(BaseTask):
         postprocess = self.model_cfg.model.decode_head
         return postprocess
 
-    def get_model_name(self) -> str:
+    def get_model_name(self, *args, **kwargs) -> str:
         """Get the model name.
 
         Return:

mmdeploy/utils/__init__.py

@@ -22,7 +22,7 @@ if importlib.util.find_spec('mmcv') is not None:
                                get_dynamic_axes, get_input_shape,
                                get_ir_config, get_model_inputs,
                                get_onnx_config, get_partition_config,
-                               get_task_type, is_dynamic_batch,
+                               get_precision, get_task_type, is_dynamic_batch,
                                is_dynamic_shape, load_config)
 
     # yapf: enable
@@ -33,5 +33,5 @@ if importlib.util.find_spec('mmcv') is not None:
         'get_codebase_config', 'get_common_config', 'get_dynamic_axes',
         'get_input_shape', 'get_ir_config', 'get_model_inputs',
         'get_onnx_config', 'get_partition_config', 'get_task_type',
-        'is_dynamic_batch', 'is_dynamic_shape', 'load_config'
+        'is_dynamic_batch', 'is_dynamic_shape', 'load_config', 'get_precision'
     ]

mmdeploy/utils/config_utils.py

@@ -380,3 +380,26 @@ def get_dynamic_axes(
                 raise KeyError('No names were found to define dynamic axes.')
         dynamic_axes = dict(zip(axes_names, dynamic_axes))
     return dynamic_axes
+
+
+def get_precision(deploy_cfg: Union[str, mmengine.Config]) -> str:
+    """Get precision of config.
+
+    Args:
+        deploy_cfg (str | mmengine.Config): The path or content of config.
+
+    Returns:
+        str: The precision of target backend.
+    """
+    precision = 'FP32'
+    deploy_cfg = load_config(deploy_cfg)[0]
+    backend = get_backend(deploy_cfg=deploy_cfg)
+    if backend == Backend.TENSORRT:
+        common_cfg = get_common_config(deploy_cfg)
+        if common_cfg.get('fp16_mode', False):
+            precision = 'FP16'
+        if common_cfg.get('int8_mode', False):
+            precision = 'INT8'
+    if backend == Backend.NCNN and 'precision' in deploy_cfg['backend_config']:
+        precision = deploy_cfg['backend_config']['precision']
+    return precision