fix_deploy_python_infer (#2233)

dygraph/deploy/python/infer.py

@@ -23,7 +23,7 @@ from PIL import Image
 import cv2
 import numpy as np
 import paddle
-from preprocess import preprocess, ResizeOp, NormalizeImageOp, PermuteOp, PadStride
+from preprocess import preprocess, Resize, NormalizeImage, Permute, PadStride
 from visualize import visualize_box_mask
 from paddle.inference import Config
 from paddle.inference import create_predictor
@@ -196,22 +196,22 @@ class DetectorSOLOv2(Detector):
             self.predictor.run()
             output_names = self.predictor.get_output_names()
             np_label = self.predictor.get_output_handle(output_names[
-                0]).copy_to_cpu()
-            np_score = self.predictor.get_output_handle(output_names[
                 1]).copy_to_cpu()
-            np_segms = self.predictor.get_output_handle(output_names[
+            np_score = self.predictor.get_output_handle(output_names[
                 2]).copy_to_cpu()
+            np_segms = self.predictor.get_output_handle(output_names[
+                3]).copy_to_cpu()
 
         t1 = time.time()
         for i in range(repeats):
             self.predictor.run()
             output_names = self.predictor.get_output_names()
             np_label = self.predictor.get_output_handle(output_names[
-                0]).copy_to_cpu()
-            np_score = self.predictor.get_output_handle(output_names[
                 1]).copy_to_cpu()
-            np_segms = self.predictor.get_output_handle(output_names[
+            np_score = self.predictor.get_output_handle(output_names[
                 2]).copy_to_cpu()
+            np_segms = self.predictor.get_output_handle(output_names[
+                3]).copy_to_cpu()
         t2 = time.time()
         ms = (t2 - t1) * 1000.0 / repeats
         print("Inference: {} ms per batch image".format(ms))

dygraph/deploy/python/preprocess.py

@@ -38,7 +38,7 @@ def decode_image(im_file, im_info):
     return im, im_info
 
 
-class ResizeOp(object):
+class Resize(object):
     """resize image by target_size and max_size
     Args:
         target_size (int): the target size of image
@@ -115,7 +115,7 @@ class ResizeOp(object):
         return im_scale_y, im_scale_x
 
 
-class NormalizeImageOp(object):
+class NormalizeImage(object):
     """normalize image
     Args:
         mean (list): im - mean
@@ -150,7 +150,7 @@ class NormalizeImageOp(object):
         return im, im_info
 
 
-class PermuteOp(object):
+class Permute(object):
     """permute image
     Args:
         to_bgr (bool): whether convert RGB to BGR 
@@ -158,7 +158,7 @@ class PermuteOp(object):
     """
 
     def __init__(self, ):
-        super(PermuteOp, self).__init__()
+        super(Permute, self).__init__()
 
     def __call__(self, im, im_info):
         """

dygraph/deploy/python/visualize.py

@@ -173,7 +173,7 @@ def draw_segm(im,
     clsid2color = {}
     np_segms = np_segms.astype(np.uint8)
     for i in range(np_segms.shape[0]):
-        mask, score, clsid = np_segms[i], np_score[i], np_label[i] + 1
+        mask, score, clsid = np_segms[i], np_score[i], np_label[i]
         if score < threshold:
             continue
 

dygraph/docs/tutorials/INSTALL_cn.md

@@ -12,7 +12,7 @@
 
 **环境需求:**
 
-- PaddlePaddle 每日版本
+- PaddlePaddle 2.0.1 或 PaddlePaddle release/2.0分支最新编译安装包
 - OS 64位操作系统
 - Python 3(3.5.1+/3.6/3.7)，64位版本
 - pip/pip3(9.0.1+)，64位版本

dygraph/ppdet/engine/export_utils.py

@@ -52,7 +52,7 @@ def _parse_reader(reader_cfg, dataset_cfg, metric, arch, image_shape):
     for st in sample_transforms[1:]:
         for key, value in st.items():
             p = {'type': key}
-            if key == 'ResizeOp':
+            if key == 'Resize':
                 if value.get('keep_ratio',
                              False) and image_shape[1] is not None:
                     max_size = max(image_shape[1:])
@@ -65,7 +65,7 @@ def _parse_reader(reader_cfg, dataset_cfg, metric, arch, image_shape):
         methods = [list(bt.keys())[0] for bt in batch_transforms]
         for bt in batch_transforms:
             for key, value in bt.items():
-                if key == 'PadBatchOp':
+                if key == 'PadBatch':
                     preprocess_list.append({'type': 'PadStride'})
                     preprocess_list[-1].update({
                         'stride': value['pad_to_stride']