update config for ShiTu_rec (#3245)

* update config for ShiTu_rec

* update for ShiTu_Rec

ppcls/configs/GeneralRecognitionV2/GeneralRecognitionV2_CLIP_vit_base.yaml

@@ -17,10 +17,14 @@ Global:
   use_dali: False
   to_static: False
 
+# mixed precision
 AMP:
-  scale_loss: 128
+  use_amp: False
+  use_fp16_test: False
+  scale_loss: 128.0
   use_dynamic_loss_scaling: True
-  # O1: mixed fp16
+  use_promote: False
+  # O1: mixed fp16, O2: pure fp16
   level: O1
 
 # model architecture

ppcls/configs/GeneralRecognitionV2/GeneralRecognitionV2_CLIP_vit_large.yaml

@@ -17,10 +17,14 @@ Global:
   use_dali: False
   to_static: False
 
+# mixed precision
 AMP:
-  scale_loss: 128
+  use_amp: False
+  use_fp16_test: False
+  scale_loss: 128.0
   use_dynamic_loss_scaling: True
-  # O1: mixed fp16
+  use_promote: False
+  # O1: mixed fp16, O2: pure fp16
   level: O1
 
 # model architecture

ppcls/configs/GeneralRecognitionV2/GeneralRecognitionV2_PPLCNetV2_base.yaml

@@ -18,10 +18,14 @@ Global:
   image_shape: [3, 224, 224]
   save_inference_dir: ./inference
 
+# mixed precision
 AMP:
-  scale_loss: 65536
+  use_amp: False
+  use_fp16_test: False
+  scale_loss: 128.0
   use_dynamic_loss_scaling: True
-  # O1: mixed fp16
+  use_promote: False
+  # O1: mixed fp16, O2: pure fp16
   level: O1
 
 # model architecture

ppcls/data/__init__.py

@@ -193,5 +193,5 @@ def build_dataloader(config, mode, device, use_dali=False, seed=None):
 
 # for PaddleX
 ClsDataset = ImageNetDataset
-ShiTuDataset = ImageNetDataset
+ShiTuRecDataset = ImageNetDataset
 MLClsDataset = MultiLabelDataset

ppcls/utils/config.py

@@ -226,7 +226,13 @@ def setup_orderdict():
 def dump_infer_config(config, path):
     setup_orderdict()
     infer_cfg = OrderedDict()
-    transforms = config["Infer"]["transforms"]
+    if config.get("Infer"):
+        transforms = config["Infer"]["transforms"]
+    elif config["DataLoader"]["Eval"].get("Query"):
+        transforms = config["DataLoader"]["Eval"]["Query"]["dataset"]["transform_ops"]
+        transforms.append({"ToCHWImage": None})
+    else:
+        logger.error("This config does not support dump transform config!")
     for transform in transforms:
         if "NormalizeImage" in transform:
             transform["NormalizeImage"]["channel_num"] = 3
@@ -241,28 +247,30 @@ def dump_infer_config(config, path):
             if "DecodeImage" not in infer_preprocess
         ]
     }
+    if config.get("Infer"):
+        postprocess_dict = config["Infer"]["PostProcess"]
 
-    postprocess_dict = config["Infer"]["PostProcess"]
-    with open(postprocess_dict["class_id_map_file"], 'r') as f:
-        label_id_maps = f.readlines()
-    label_names = []
-    for line in label_id_maps:
-        line = line.strip().split(' ', 1)
-        label_names.append(line[1:][0])
+        with open(postprocess_dict["class_id_map_file"], 'r') as f:
+            label_id_maps = f.readlines()
+        label_names = []
+        for line in label_id_maps:
+            line = line.strip().split(' ', 1)
+            label_names.append(line[1:][0])
 
-    postprocess_name = postprocess_dict.get("name", None)
-    postprocess_dict.pop("class_id_map_file")
-    postprocess_dict.pop("name")
-    dic = OrderedDict()
-    for item in postprocess_dict.items():
-        dic[item[0]] = item[1]
-    dic['label_list'] = label_names
+        postprocess_name = postprocess_dict.get("name", None)
+        postprocess_dict.pop("class_id_map_file")
+        postprocess_dict.pop("name")
+        dic = OrderedDict()
+        for item in postprocess_dict.items():
+           dic[item[0]] = item[1]
+        dic['label_list'] = label_names
 
-    if postprocess_name:
-        infer_cfg["PostProcess"] = {postprocess_name: dic}
+        if postprocess_name:
+            infer_cfg["PostProcess"] = {postprocess_name: dic}
+        else:
+            raise ValueError("PostProcess name is not specified")
     else:
-        raise ValueError("PostProcess name is not specified")
-
+        infer_cfg["PostProcess"] = {"NormalizeFeatures": None}
     with open(path, 'w') as f:
         yaml.dump(infer_cfg, f)
     logger.info("Export inference config file to {}".format(os.path.join(path)))