Merge pull request #1750 from weisy11/deploy_v2

update format

deploy/python/ppshitu_v2/configs/test_config.yml

-Global:
-  Engine: POPEngine
-  infer_imgs: "../../images/wangzai.jpg"
-
-AlgoModule:
-  - Module:
-      preprocess:
-        name: ImageProcessor
+Modules:
+  - name: Detector
+    type: AlgoMod
     processors:
+      - name: ImageProcessor
+        type: preprocessor
+        ops:
           - ResizeImage:
               size: [640, 640]
               interpolation: 2
@@ -17,14 +15,16 @@ AlgoModule:
               order: hwc
           - ToCHWImage:
           - GetShapeInfo:
+              configs:
                 order: chw
           - ToBatch:
-      predictor:
+      - name: PaddlePredictor
+        type: predictor
         inference_model_dir: ./models/ppyolov2_r50vd_dcn_mainbody_v1.0_infer/
         input_names:
         output_names:
-      postprocess:
-        name: DetPostProcessor
+      - name: DetPostPro
+        type: postprocessor
         threshold: 0.2
         max_det_results: 1
         label_list:

deploy/python/ppshitu_v2/engine/pop_engine.py

+import importlib
+
 from processor.algo_mod import AlgoMod
 
 
 class POPEngine:
     def __init__(self, config):
         self.algo_list = []
-        # last_algo_type = "start"
-        for algo_config in config["AlgoModule"]:
-            # algo_config["last_algo_type"] = last_algo_type
-            self.algo_list.append(AlgoMod(algo_config["Module"]))
-            # last_algo_type = algo_config["type"]
+        current_mod = importlib.import_module(__name__)
+        for mod_config in config["Modules"]:
+            mod_type = mod_config.get("type")
+            mod = getattr(current_mod, mod_type)(mod_config)
+            self.algo_list.append(mod)
 
-    def process(self, x):
+    def process(self, input_data):
         for algo_module in self.algo_list:
-            x = algo_module.process(x)
-        return x
+            input_data = algo_module.process(input_data)
+        return input_data

deploy/python/ppshitu_v2/examples/predict.py

@@ -7,7 +7,6 @@ import cv2
 
 from engine import build_engine
 from utils import config
-from utils.get_image_list import get_image_list
 
 
 def main():
@@ -16,9 +15,7 @@ def main():
         args.config, overrides=args.override, show=False)
     config_dict.profiler_options = args.profiler_options
     engine = build_engine(config_dict)
-
-    image_list = get_image_list(config_dict["Global"]["infer_imgs"])
-    for idx, image_file in enumerate(image_list):
+    image_file = "../../images/wangzai.jpg"
     img = cv2.imread(image_file)[:, :, ::-1]
     input_data = {"input_image": img}
     output = engine.process(input_data)

deploy/python/ppshitu_v2/processor/__init__.py

-from abc import ABC, abstractmethod
-
-from processor.algo_mod import searcher
-from processor.algo_mod.predictors import build_predictor
-
-# def build_processor(config):
-#     print(config)
-#     processor_type = config.get("processor_type")
-#     processor_mod = locals()[processor_type]
-#     processor_name = config.get("processor_name")
-#     return getattr(processor_mod, processor_name)
-
-# class BaseProcessor(ABC):
-#     @abstractmethod
-#     def __init__(self, config):
-#         pass
-
-#     @abstractmethod
-#     def process(self, input_data):
-#         pass
+from .algo_mod import AlgoMod

deploy/python/ppshitu_v2/processor/algo_mod/__init__.py

-from processor.algo_mod.data_processor import ImageProcessor
-from processor.algo_mod.post_processor.det import DetPostProcessor
-from processor.algo_mod.predictors import build_predictor
+from .postprocessor import build_postprocessor
+from .preprocessor import build_preprocessor
+from .predictor import build_predictor
 
+from ..base_processor import BaseProcessor
 
-def build_processor(config):
-    # processor_type = config.get("processor_type")
-    # processor_mod = locals()[processor_type]
-    processor_name = config.get("name")
-    return eval(processor_name)(config)
 
-
-class AlgoMod(object):
+class AlgoMod(BaseProcessor):
     def __init__(self, config):
-        self.pre_processor = build_processor(config["preprocess"])
-        self.predictor = build_predictor(config["predictor"])
-        self.post_processor = build_processor(config["postprocess"])
+        self.processors = []
+        for processor_config in config["processors"]:
+            processor_type = processor_config.get("type")
+            if processor_type == "preprocessor":
+                processor = build_preprocessor(processor_config)
+            elif processor_type == "predictor":
+                processor = build_predictor(processor_config)
+            elif processor_type == "postprocessor":
+                processor = build_postprocessor(processor_config)
+            else:
+                raise NotImplemented("processor type {} unknown.".format(processor_type))
+            self.processors.append(processor)
 
     def process(self, input_data):
-        input_data = self.pre_processor.process(input_data)
-        input_data = self.predictor.process(input_data)
-        input_data = self.post_processor.process(input_data)
+        for processor in self.processors:
+            input_data = processor.process(input_data)
         return input_data

deploy/python/ppshitu_v2/processor/algo_mod/data_processor/__init__.py

-from processor.algo_mod.data_processor.image_processor import ImageProcessor
-from processor.algo_mod.data_processor.bbox_cropper import BBoxCropper

deploy/python/ppshitu_v2/processor/algo_mod/postprocessor/__init__.py

+import importlib
+
+from .det import DetPostPro
+
+
+def build_postprocessor(config):
+    processor_mod = importlib.import_module(__name__)
+    processor_name = config.get("name")
+    return getattr(processor_mod, processor_name)(config)

deploy/python/ppshitu_v2/processor/algo_mod/post_processor/det.py → deploy/python/ppshitu_v2/processor/algo_mod/postprocessor/det.py

 from functools import reduce
-
 import numpy as np
 
+from utils import logger
+from ...base_processor import BaseProcessor
+
 
-class DetPostProcessor(object):
+class DetPostPro(BaseProcessor):
     def __init__(self, config):
-        super().__init__()
         self.threshold = config["threshold"]
         self.label_list = config["label_list"]
         self.max_det_results = config["max_det_results"]
 
-    def process(self, pred):
-        np_boxes = pred["save_infer_model/scale_0.tmp_1"]
+    def process(self, input_data):
+        pred = input_data["pred"]
+        np_boxes = pred[list(pred.keys())[0]]
         if reduce(lambda x, y: x * y, np_boxes.shape) < 6:
-            print('[WARNNING] No object detected.')
+            logger.warning('[Detector] No object detected.')
             np_boxes = np.array([])
 
         keep_indexes = np_boxes[:, 1].argsort()[::-1][:self.max_det_results]

deploy/python/ppshitu_v2/processor/algo_mod/predictor/__init__.py

+import importlib
+
+from processor.algo_mod.predictor.paddle_predictor import PaddlePredictor
+from processor.algo_mod.predictor.onnx_predictor import ONNXPredictor
+
+
+def build_predictor(config):
+    processor_mod = importlib.import_module(__name__)
+    processor_name = config.get("name")
+    return getattr(processor_mod, processor_name)(config)

deploy/python/ppshitu_v2/processor/algo_mod/predictor/onnx_predictor.py

+from ...base_processor import BaseProcessor
+
+
+class ONNXPredictor(BaseProcessor):
+    def __init__(self, config):
+        pass
+    def process(self, input_data):
+        raise NotImplemented("ONNXPredictor Not supported yet")

deploy/python/ppshitu_v2/processor/algo_mod/predictors/paddle_predictor.py → deploy/python/ppshitu_v2/processor/algo_mod/predictor/paddle_predictor.py

 import os
 import platform
-
 from paddle.inference import create_predictor
 from paddle.inference import Config as PaddleConfig
 
+from ...base_processor import BaseProcessor
+
 
-class Predictor(object):
+class PaddlePredictor(BaseProcessor):
     def __init__(self, config):
-        super().__init__()
-        # HALF precission predict only work when using tensorrt
         if config.get("use_fp16", False):
             assert config.get("use_tensorrt", False) is True
 
@@ -61,5 +60,5 @@ class Predictor(object):
         for output_name in output_names:
             output = self.predictor.get_output_handle(output_name)
             output_data[output_name] = output.copy_to_cpu()
-
-        return output_data
+        input_data["pred"] = output_data
+        return input_data

deploy/python/ppshitu_v2/processor/algo_mod/predictors/__init__.py

-from processor.algo_mod.predictors.paddle_predictor import Predictor as paddle_predictor
-from processor.algo_mod.predictors.onnx_predictor import Predictor as onnx_predictor
-
-
-def build_predictor(config):
-    # if use paddle backend
-    if True:
-        return paddle_predictor(config)
-    # if use onnx backend
-    else:
-        return onnx_predictor(config)
\ No newline at end of file

deploy/python/ppshitu_v2/processor/algo_mod/predictors/onnx_predictor.py

-class Predictor(object):
-    def __init__(self, config):
-        super().__init__()

deploy/python/ppshitu_v2/processor/algo_mod/preprocessor/__init__.py

+import importlib
+
+from processor.algo_mod.preprocessor.image_processor import ImageProcessor
+
+
+def build_preprocessor(config):
+    processor_mod = importlib.import_module(__name__)
+    processor_name = config.get("name")
+    return getattr(processor_mod, processor_name)(config)

deploy/python/ppshitu_v2/processor/algo_mod/data_processor/image_processor.py → deploy/python/ppshitu_v2/processor/algo_mod/preprocessor/image_processor.py

@@ -3,32 +3,20 @@ import cv2
 import numpy as np
 import importlib
 from PIL import Image
-import paddle
 
 from utils import logger
-# from processor import BaseProcessor
+from processor.base_processor import BaseProcessor
 
-from abc import ABC, abstractmethod
 
-
-class BaseProcessor(ABC):
-    @abstractmethod
-    def __init__(self, *args, **kwargs):
-        pass
-
-    @abstractmethod
-    def process(self, input_data):
-        pass
-
-
-class ImageProcessor(object):
+class ImageProcessor(BaseProcessor):
     def __init__(self, config):
         self.processors = []
-        for processor_config in config.get("processors"):
+        mod = importlib.import_module(__name__)
+        for processor_config in config.get("ops"):
             name = list(processor_config)[0]
             param = {} if processor_config[name] is None else processor_config[
                 name]
-            op = eval(name)(**param)
+            op = getattr(mod, name)(**param)
             self.processors.append(op)
 
     def process(self, input_data):
@@ -39,13 +27,13 @@ class ImageProcessor(object):
                 input_data = processor.process(input_data)
             else:
                 image = processor(image)
+        input_data["image"] = image
         return input_data
 
 
 class GetShapeInfo(BaseProcessor):
-    def __init__(self, order="hwc"):
-        super().__init__()
-        self.order = order
+    def __init__(self, configs):
+        self.order = configs.get("order")
 
     def process(self, input_data):
         input_image = input_data["input_image"]
@@ -69,43 +57,22 @@ class GetShapeInfo(BaseProcessor):
                 ],
                 dtype=np.float32)
         input_data['input_shape'] = np.array(image.shape[:2], dtype=np.float32)
-        print(image.shape[0])
         return input_data
 
 
-# class ToTensor(BaseProcessor):
-#     def __init__(self):
-#         super().__init__()
-
-#     def process(self, input_data):
-#         image = input_data["image"]
-#         input_data["input_tensor"] = paddle.to_tensor(image)
-#         return input_data
-
-
-class ToBatch(BaseProcessor):
-    def __init__(self):
-        super().__init__()
-
-    def process(self, input_data):
-        image = input_data["image"]
-        input_data["image"] = image[np.newaxis, :, :, :]
-        return input_data
+class ToBatch:
+    def __call__(self, img):
+        img = img[np.newaxis, :, :, :]
+        return img
 
 
 class ToRGB:
-    def __init__(self):
-        pass
-
     def __call__(self, img):
         img = img[:, :, ::-1]
         return img
 
 
 class ToCHWImage:
-    def __init__(self):
-        pass
-
     def __call__(self, img, img_info=None):
         img = img.transpose((2, 0, 1))
         return img

deploy/python/ppshitu_v2/processor/algo_mod/data_processor/bbox_cropper.py → deploy/python/ppshitu_v2/processor/base_processor.py

-from processor.algo_mod.data_processor.image_processor import BaseProcessor
+from abc import ABC, abstractmethod
 
 
-class BBoxCropper(BaseProcessor):
+class BaseProcessor(ABC):
+    @abstractmethod
     def __init__(self, config):
         pass
 
+    @abstractmethod
     def process(self, input_data):
         pass