diff --git a/ppcls/configs/PULC/safety_helmet/MobileNetV3_small_x0_35.yaml b/ppcls/configs/PULC/safety_helmet/MobileNetV3_small_x0_35.yaml
index c9a16384e19981b6941be8268c88ace4fb38ae23..2b536e1b99cf9656c8ed141c50e62bfee6bfc762 100644
--- a/ppcls/configs/PULC/safety_helmet/MobileNetV3_small_x0_35.yaml
+++ b/ppcls/configs/PULC/safety_helmet/MobileNetV3_small_x0_35.yaml
@@ -7,7 +7,7 @@ Global:
   save_interval: 1
   eval_during_train: True
   eval_interval: 1
-  epochs: 40
+  epochs: 60
   print_batch_step: 10
   use_visualdl: False
   # used for static mode and model export
diff --git a/ppcls/configs/PULC/safety_helmet/PPLCNet_x1_0.yaml b/ppcls/configs/PULC/safety_helmet/PPLCNet_x1_0.yaml
index 3ad59e2e0078dcf43a273744f559d05ede567ab3..4c3c8642d9464025dda4d628d328e34d3b8a1613 100644
--- a/ppcls/configs/PULC/safety_helmet/PPLCNet_x1_0.yaml
+++ b/ppcls/configs/PULC/safety_helmet/PPLCNet_x1_0.yaml
@@ -13,12 +13,14 @@ Global:
   # used for static mode and model export
   image_shape: [3, 224, 224]
   save_inference_dir: ./inference
+
 # model architecture
 Arch:
   name: PPLCNet_x1_0
   pretrained: True
   use_ssld: True
   class_num: 2
+  use_sync_bn : True
  
 # loss function config for traing/eval process
 Loss:
@@ -36,7 +38,7 @@ Optimizer:
   momentum: 0.9
   lr:
     name: Cosine
-    learning_rate: 0.01
+    learning_rate: 0.025
     warmup_epoch: 5
   regularizer:
     name: 'L2'
@@ -55,15 +57,27 @@ DataLoader:
             to_rgb: True
             channel_first: False
         - RandCropImage:
-            size: 224
+            size: 176
         - RandFlipImage:
             flip_code: 1
+        - TimmAutoAugment:
+            prob : 0.5
+            config_str: rand-m9-mstd0.5-inc1
+            interpolation: bicubic
+            img_size : 176
         - NormalizeImage:
             scale: 1.0/255.0
             mean: [0.485, 0.456, 0.406]
             std: [0.229, 0.224, 0.225]
             order: ''
-
+        - RandomErasing:
+            EPSILON : 0.1
+            r1 : 0.3
+            sh : 1.0/3.0
+            sl : 0.02
+            attempt : 10
+            use_log_aspect : True
+            mode : pixel
     sampler:
       name: DistributedBatchSampler
       batch_size: 64
diff --git a/ppcls/configs/PULC/safety_helmet/PPLCNet_x1_0_search.yaml b/ppcls/configs/PULC/safety_helmet/PPLCNet_x1_0_search.yaml
index 58cc3300defe138ade905516a9a5b7e88e2a12bf..98f63a613278cf7eed879d60a65e942bdfb4c687 100644
--- a/ppcls/configs/PULC/safety_helmet/PPLCNet_x1_0_search.yaml
+++ b/ppcls/configs/PULC/safety_helmet/PPLCNet_x1_0_search.yaml
@@ -37,7 +37,7 @@ Optimizer:
   momentum: 0.9
   lr:
     name: Cosine
-    learning_rate: 0.015
+    learning_rate: 0.10
     warmup_epoch: 5
   regularizer:
     name: 'L2'
@@ -60,7 +60,7 @@ DataLoader:
         - RandFlipImage:
             flip_code: 1
         - TimmAutoAugment:
-            prob: 0.5
+            prob: 0
             config_str: rand-m9-mstd0.5-inc1
             interpolation: bicubic
             img_size: 192
@@ -70,7 +70,7 @@ DataLoader:
             std: [0.229, 0.224, 0.225]
             order: ''
         - RandomErasing:
-            EPSILON: 0.5
+            EPSILON: 0
             sl: 0.02
             sh: 1.0/3.0
             r1: 0.3
diff --git a/ppcls/configs/PULC/safety_helmet/Res2Net200_vd_26w_4s.yaml b/ppcls/configs/PULC/safety_helmet/Res2Net200_vd_26w_4s.yaml
index 5ab2f484e6f54c44ca5c5c4b40947afb0b6160a0..5b987d510c17df28c100728690f8a5d62293d36c 100644
--- a/ppcls/configs/PULC/safety_helmet/Res2Net200_vd_26w_4s.yaml
+++ b/ppcls/configs/PULC/safety_helmet/Res2Net200_vd_26w_4s.yaml
@@ -7,7 +7,7 @@ Global:
   save_interval: 1
   eval_during_train: True
   eval_interval: 1
-  epochs: 40
+  epochs: 60
   print_batch_step: 10
   use_visualdl: False
   # used for static mode and model export
diff --git a/ppcls/configs/PULC/safety_helmet/SwinTransformer_tiny_patch4_window7_224.yaml b/ppcls/configs/PULC/safety_helmet/SwinTransformer_tiny_patch4_window7_224.yaml
index 666ad6be28d432c4bd5a831f7eb8ddd1ccb269dd..5863ee17e4627cff71444710988e94ae76cd8025 100644
--- a/ppcls/configs/PULC/safety_helmet/SwinTransformer_tiny_patch4_window7_224.yaml
+++ b/ppcls/configs/PULC/safety_helmet/SwinTransformer_tiny_patch4_window7_224.yaml
@@ -7,7 +7,7 @@ Global:
   save_interval: 1
   eval_during_train: True
   eval_interval: 1
-  epochs: 40
+  epochs: 60
   print_batch_step: 10
   use_visualdl: False
   # used for static mode and model export
@@ -81,6 +81,14 @@ DataLoader:
             attempt: 10
             use_log_aspect: True
             mode: pixel
+      batch_transform_ops:
+        - OpSampler:
+            MixupOperator:
+              alpha: 0.8
+              prob: 0.5
+            CutmixOperator:
+              alpha: 1.0
+              prob: 0.5
 
     sampler:
       name: DistributedBatchSampler
diff --git a/ppcls/configs/PULC/safety_helmet/search.yaml b/ppcls/configs/PULC/safety_helmet/search.yaml
index 09bf2e9e5cdfe0008a2c0010a29d6a33a8d879a1..e8c1c933dfef09f1d3a6189f4701a1e7d0678ab9 100644
--- a/ppcls/configs/PULC/safety_helmet/search.yaml
+++ b/ppcls/configs/PULC/safety_helmet/search.yaml
@@ -8,7 +8,7 @@ search_dict:
   - search_key: lrs
     replace_config:
       - Optimizer.lr.learning_rate
-    search_values: [0.01, 0.015, 0.025]
+    search_values: [0.03, 0.04, 0.05, 0.06, 0.07, 0.08, 0.09, 0.10, 0.11, 0.11, 0.12]
   - search_key: resolutions
     replace_config:
       - DataLoader.Train.dataset.transform_ops.1.RandCropImage.size