diff --git a/contrib/ACE2P/config.py b/contrib/ACE2P/config.py
index f6ad509581a84d04bc1b6badca83648505c19444..a1fad0ec1e6c50493a0a8dfab0c5301add410ad0 100644
--- a/contrib/ACE2P/config.py
+++ b/contrib/ACE2P/config.py
@@ -6,13 +6,13 @@ args = get_arguments()
 cfg = AttrDict()
 
 # 待预测图像所在路径
-cfg.data_dir = os.path.join(args.example , "data", "testing_images")
+cfg.data_dir = os.path.join("data", "testing_images")
 # 待预测图像名称列表
-cfg.data_list_file = os.path.join(args.example , "data", "test_id.txt")
+cfg.data_list_file = os.path.join("data", "test_id.txt")
 # 模型加载路径
-cfg.model_path = os.path.join(args.example , "ACE2P")
+cfg.model_path = args.example
 # 预测结果保存路径
-cfg.vis_dir = os.path.join(args.example , "result")
+cfg.vis_dir = "result"
 
 # 预测类别数
 cfg.class_num = 20
diff --git a/contrib/ACE2P/download_ACE2P.py b/contrib/ACE2P/download_ACE2P.py
new file mode 100644
index 0000000000000000000000000000000000000000..bb4d33771dbd879a2d77664d2e0e45ed33b9bcb2
--- /dev/null
+++ b/contrib/ACE2P/download_ACE2P.py
@@ -0,0 +1,31 @@
+# Copyright (c) 2019  PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import sys
+import os
+
+LOCAL_PATH = os.path.dirname(os.path.abspath(__file__))
+TEST_PATH = os.path.join(LOCAL_PATH, "..", "..", "test")
+sys.path.append(TEST_PATH)
+
+from test_utils import download_file_and_uncompress
+
+if __name__ == "__main__":
+    download_file_and_uncompress(
+        url='https://paddleseg.bj.bcebos.com/models/ACE2P.tgz',
+        savepath=LOCAL_PATH,
+        extrapath=LOCAL_PATH,
+        extraname='ACE2P')
+
+    print("Pretrained Model download success!")
diff --git a/contrib/ACE2P/imgs/117676_2149260.jpg b/contrib/ACE2P/imgs/117676_2149260.jpg
new file mode 100644
index 0000000000000000000000000000000000000000..8314d8f8cc723b6f96785053bdcfe39d867755d5
Binary files /dev/null and b/contrib/ACE2P/imgs/117676_2149260.jpg differ
diff --git a/contrib/ACE2P/imgs/117676_2149260.png b/contrib/ACE2P/imgs/117676_2149260.png
new file mode 100644
index 0000000000000000000000000000000000000000..e3a9529644ead2013748431a3ade2f34264f19de
Binary files /dev/null and b/contrib/ACE2P/imgs/117676_2149260.png differ
diff --git a/contrib/ACE2P/infer.py b/contrib/ACE2P/infer.py
new file mode 100644
index 0000000000000000000000000000000000000000..16eddc1eab8628eec7e38d27b1f18df13dd480d7
--- /dev/null
+++ b/contrib/ACE2P/infer.py
@@ -0,0 +1,130 @@
+# -*- coding: utf-8 -*-
+import os
+import cv2
+import numpy as np
+from utils.util import get_arguments
+from utils.palette import get_palette
+from PIL import Image as PILImage
+import importlib
+
+args = get_arguments()
+config = importlib.import_module('config')
+cfg = getattr(config, 'cfg')
+
+# paddle垃圾回收策略FLAG，ACE2P模型较大，当显存不够时建议开启
+os.environ['FLAGS_eager_delete_tensor_gb']='0.0'
+
+import paddle.fluid as fluid
+
+# 预测数据集类
+class TestDataSet():
+    def __init__(self):
+        self.data_dir = cfg.data_dir 
+        self.data_list_file = cfg.data_list_file
+        self.data_list = self.get_data_list()
+        self.data_num = len(self.data_list)
+    
+    def get_data_list(self):
+        # 获取预测图像路径列表
+        data_list = []
+        data_file_handler = open(self.data_list_file, 'r')
+        for line in data_file_handler:
+            img_name = line.strip()
+            name_prefix = img_name.split('.')[0]
+            if len(img_name.split('.')) == 1:
+                img_name = img_name + '.jpg'
+            img_path = os.path.join(self.data_dir, img_name)
+            data_list.append(img_path)
+        return data_list
+
+    def preprocess(self, img):
+        # 图像预处理
+        if cfg.example == 'ACE2P':
+            reader = importlib.import_module('reader')
+            ACE2P_preprocess = getattr(reader, 'preprocess')
+            img = ACE2P_preprocess(img)
+        else:
+            img = cv2.resize(img, cfg.input_size).astype(np.float32)
+            img -= np.array(cfg.MEAN)
+            img /= np.array(cfg.STD)
+            img = img.transpose((2, 0, 1))
+            img = np.expand_dims(img, axis=0)
+        return img
+
+    def get_data(self, index):
+        # 获取图像信息
+        img_path = self.data_list[index]
+        img = cv2.imread(img_path, cv2.IMREAD_COLOR)
+        if img is None:
+            return img, img,img_path, None
+
+        img_name = img_path.split(os.sep)[-1]
+        name_prefix = img_name.replace('.'+img_name.split('.')[-1],'')
+        img_shape = img.shape[:2]
+        img_process = self.preprocess(img)
+
+        return img, img_process, name_prefix, img_shape
+
+
+def infer():
+    if not os.path.exists(cfg.vis_dir):
+        os.makedirs(cfg.vis_dir)
+    palette = get_palette(cfg.class_num)
+    # 人像分割结果显示阈值
+    thresh = 120
+
+    place = fluid.CUDAPlace(0) if cfg.use_gpu else fluid.CPUPlace()
+    exe = fluid.Executor(place)
+
+    # 加载预测模型
+    test_prog, feed_name, fetch_list = fluid.io.load_inference_model(
+        dirname=cfg.model_path, executor=exe, params_filename='__params__')
+
+    #加载预测数据集
+    test_dataset = TestDataSet()
+    data_num = test_dataset.data_num
+
+    for idx in range(data_num):
+        # 数据获取
+        ori_img, image, im_name, im_shape = test_dataset.get_data(idx)
+        if image is None:
+            print(im_name, 'is None')
+            continue
+        
+        # 预测
+        if cfg.example == 'ACE2P':
+            # ACE2P模型使用多尺度预测
+            reader = importlib.import_module('reader')
+            multi_scale_test = getattr(reader, 'multi_scale_test')
+            parsing, logits = multi_scale_test(exe, test_prog, feed_name, fetch_list, image, im_shape)
+        else:
+            # HumanSeg,RoadLine模型单尺度预测
+            result = exe.run(program=test_prog, feed={feed_name[0]: image}, fetch_list=fetch_list)
+            parsing = np.argmax(result[0][0], axis=0)
+            parsing = cv2.resize(parsing.astype(np.uint8), im_shape[::-1])
+        
+        # 预测结果保存
+        result_path = os.path.join(cfg.vis_dir, im_name + '.png')
+        if cfg.example == 'HumanSeg':
+            logits = result[0][0][1]*255
+            logits = cv2.resize(logits, im_shape[::-1])
+            ret, logits = cv2.threshold(logits, thresh, 0, cv2.THRESH_TOZERO)
+            logits = 255 *(logits - thresh)/(255 - thresh)
+            # 将分割结果添加到alpha通道
+            rgba = np.concatenate((ori_img, np.expand_dims(logits, axis=2)), axis=2)
+            cv2.imwrite(result_path, rgba)
+        else: 
+            output_im = PILImage.fromarray(np.asarray(parsing, dtype=np.uint8))
+            output_im.putpalette(palette)
+            output_im.save(result_path)
+
+        if (idx + 1) % 100 == 0:
+            print('%d  processd' % (idx + 1))
+            
+    print('%d  processd done' % (idx + 1))   
+    
+    return 0
+
+
+if __name__ == "__main__":
+    infer()
diff --git a/contrib/ACE2P/reader.py b/contrib/ACE2P/reader.py
index 0a266637f3cf425a1bc3d61ad7377ff30de55723..ef5cc370738daf8550adfc20c227f942f1dd300f 100644
--- a/contrib/ACE2P/reader.py
+++ b/contrib/ACE2P/reader.py
@@ -1,7 +1,7 @@
 # -*- coding: utf-8 -*- 
 import numpy as np
 import paddle.fluid as fluid
-from ACE2P.config import cfg
+from config import cfg
 import cv2
 
 def get_affine_points(src_shape, dst_shape, rot_grad=0):
diff --git a/contrib/utils/__init__.py b/contrib/ACE2P/utils/__init__.py
similarity index 100%
rename from contrib/utils/__init__.py
rename to contrib/ACE2P/utils/__init__.py
diff --git a/contrib/utils/palette.py b/contrib/ACE2P/utils/palette.py
similarity index 100%
rename from contrib/utils/palette.py
rename to contrib/ACE2P/utils/palette.py
diff --git a/contrib/utils/util.py b/contrib/ACE2P/utils/util.py
similarity index 100%
rename from contrib/utils/util.py
rename to contrib/ACE2P/utils/util.py
diff --git a/contrib/imgs/Human.jpg b/contrib/HumanSeg/imgs/Human.jpg
similarity index 100%
rename from contrib/imgs/Human.jpg
rename to contrib/HumanSeg/imgs/Human.jpg
diff --git a/contrib/imgs/HumanSeg.jpg b/contrib/HumanSeg/imgs/HumanSeg.jpg
similarity index 100%
rename from contrib/imgs/HumanSeg.jpg
rename to contrib/HumanSeg/imgs/HumanSeg.jpg
diff --git a/contrib/infer.py b/contrib/HumanSeg/infer.py
similarity index 98%
rename from contrib/infer.py
rename to contrib/HumanSeg/infer.py
index 8f939c8455cd3868120781a7a8d96ace0ff772b1..971476933c431977ce80c73e1d939fe079e1af19 100644
--- a/contrib/infer.py
+++ b/contrib/HumanSeg/infer.py
@@ -8,7 +8,7 @@ from PIL import Image as PILImage
 import importlib
 
 args = get_arguments()
-config = importlib.import_module(args.example+'.config')
+config = importlib.import_module('config')
 cfg = getattr(config, 'cfg')
 
 # paddle垃圾回收策略FLAG，ACE2P模型较大，当显存不够时建议开启
diff --git a/contrib/HumanSeg/utils/__init__.py b/contrib/HumanSeg/utils/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/contrib/HumanSeg/utils/palette.py b/contrib/HumanSeg/utils/palette.py
new file mode 100644
index 0000000000000000000000000000000000000000..2186203cbc2789f6eff70dfd92f724b4fe16cdb7
--- /dev/null
+++ b/contrib/HumanSeg/utils/palette.py
@@ -0,0 +1,38 @@
+##+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+## Created by: RainbowSecret
+## Microsoft Research
+## yuyua@microsoft.com
+## Copyright (c) 2018
+##
+## This source code is licensed under the MIT-style license found in the
+## LICENSE file in the root directory of this source tree
+##+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+import numpy as np
+import cv2
+
+
+def get_palette(num_cls):
+    """ Returns the color map for visualizing the segmentation mask.
+    Args:
+        num_cls: Number of classes
+    Returns:
+        The color map
+    """
+    n = num_cls
+    palette = [0] * (n * 3)
+    for j in range(0, n):
+        lab = j
+        palette[j * 3 + 0] = 0
+        palette[j * 3 + 1] = 0
+        palette[j * 3 + 2] = 0
+        i = 0
+        while lab:
+            palette[j * 3 + 0] |= (((lab >> 0) & 1) << (7 - i))
+            palette[j * 3 + 1] |= (((lab >> 1) & 1) << (7 - i))
+            palette[j * 3 + 2] |= (((lab >> 2) & 1) << (7 - i))
+            i += 1
+            lab >>= 3
+    return palette
diff --git a/contrib/HumanSeg/utils/util.py b/contrib/HumanSeg/utils/util.py
new file mode 100644
index 0000000000000000000000000000000000000000..7394870e7c94c1fb16169e314696b931eecdc3b2
--- /dev/null
+++ b/contrib/HumanSeg/utils/util.py
@@ -0,0 +1,47 @@
+from __future__ import division
+from __future__ import print_function
+from __future__ import unicode_literals
+import argparse
+import os
+
+def get_arguments():
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--use_gpu",
+                        action="store_true",
+                        help="Use gpu or cpu to test.")
+    parser.add_argument('--example',
+                        type=str,
+                        help='RoadLine, HumanSeg or ACE2P')
+                        
+    return parser.parse_args()
+
+
+class AttrDict(dict):
+    def __init__(self, *args, **kwargs):
+        super(AttrDict, self).__init__(*args, **kwargs)
+
+    def __getattr__(self, name):
+        if name in self.__dict__:
+            return self.__dict__[name]
+        elif name in self:
+            return self[name]
+        else:
+            raise AttributeError(name)
+
+    def __setattr__(self, name, value):
+        if name in self.__dict__:
+            self.__dict__[name] = value
+        else:
+            self[name] = value
+
+def merge_cfg_from_args(args, cfg):
+    """Merge config keys, values in args into the global config."""
+    for k, v in vars(args).items():
+        d = cfg
+        try:
+            value = eval(v)
+        except:
+            value = v
+        if value is not None:
+            cfg[k] = value
+
diff --git a/dataset/download_mini_mechanical_industry_meter.py b/contrib/MechanicalIndustryMeter/download_mini_mechanical_industry_meter.py
similarity index 95%
rename from dataset/download_mini_mechanical_industry_meter.py
rename to contrib/MechanicalIndustryMeter/download_mini_mechanical_industry_meter.py
index 3049df25219df7641990cedd409566779012a08d..f0409581ea9454417c545aa616b98ee8ece4dc53 100644
--- a/dataset/download_mini_mechanical_industry_meter.py
+++ b/contrib/MechanicalIndustryMeter/download_mini_mechanical_industry_meter.py
@@ -16,7 +16,7 @@ import sys
 import os
 
 LOCAL_PATH = os.path.dirname(os.path.abspath(__file__))
-TEST_PATH = os.path.join(LOCAL_PATH, "..", "test")
+TEST_PATH = os.path.join(LOCAL_PATH, "..", "..", "test")
 sys.path.append(TEST_PATH)
 
 from test_utils import download_file_and_uncompress
diff --git a/contrib/MechanicalIndustryMeter/download_unet_mechanical_industry_meter.py b/contrib/MechanicalIndustryMeter/download_unet_mechanical_industry_meter.py
new file mode 100644
index 0000000000000000000000000000000000000000..aa55bf5e03b8dcf31e52043fd5dc87086c03c32f
--- /dev/null
+++ b/contrib/MechanicalIndustryMeter/download_unet_mechanical_industry_meter.py
@@ -0,0 +1,30 @@
+# Copyright (c) 2019  PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import sys
+import os
+
+LOCAL_PATH = os.path.dirname(os.path.abspath(__file__))
+TEST_PATH = os.path.join(LOCAL_PATH, "..", "..", "test")
+sys.path.append(TEST_PATH)
+
+from test_utils import download_file_and_uncompress
+
+if __name__ == "__main__":
+    download_file_and_uncompress(
+        url='https://paddleseg.bj.bcebos.com/models/unet_mechanical_industry_meter.tar',
+        savepath=LOCAL_PATH,
+        extrapath=LOCAL_PATH)
+
+    print("Pretrained Model download success!")
diff --git a/contrib/imgs/1560143028.5_IMG_3091.JPG b/contrib/MechanicalIndustryMeter/imgs/1560143028.5_IMG_3091.JPG
similarity index 100%
rename from contrib/imgs/1560143028.5_IMG_3091.JPG
rename to contrib/MechanicalIndustryMeter/imgs/1560143028.5_IMG_3091.JPG
diff --git a/contrib/imgs/1560143028.5_IMG_3091.png b/contrib/MechanicalIndustryMeter/imgs/1560143028.5_IMG_3091.png
similarity index 100%
rename from contrib/imgs/1560143028.5_IMG_3091.png
rename to contrib/MechanicalIndustryMeter/imgs/1560143028.5_IMG_3091.png
diff --git a/configs/unet_mechanical_meter.yaml b/contrib/MechanicalIndustryMeter/unet_mechanical_meter.yaml
similarity index 77%
rename from configs/unet_mechanical_meter.yaml
rename to contrib/MechanicalIndustryMeter/unet_mechanical_meter.yaml
index e1bc3a1183d2b435c84ad7b16002a3f604cf85b0..45ac8616f7993e15d3d262dc0e27f67624957e2a 100644
--- a/configs/unet_mechanical_meter.yaml
+++ b/contrib/MechanicalIndustryMeter/unet_mechanical_meter.yaml
@@ -21,14 +21,14 @@ DATALOADER:
     BUF_SIZE: 256
     NUM_WORKERS: 4
 DATASET: 
-    DATA_DIR: "./dataset/mini_mechanical_industry_meter_data/"
+    DATA_DIR: "./contrib/MechanicalIndustryMeter/mini_mechanical_industry_meter_data/"
     IMAGE_TYPE: "rgb"  # choice rgb or rgba
     NUM_CLASSES: 5
-    TEST_FILE_LIST: "./dataset/mini_mechanical_industry_meter_data/val_mini.txt"
+    TEST_FILE_LIST: "./contrib/MechanicalIndustryMeter/mini_mechanical_industry_meter_data/val_mini.txt"
     TEST_TOTAL_IMAGES: 8
-    TRAIN_FILE_LIST: "./dataset/mini_mechanical_industry_meter_data/train_mini.txt"
+    TRAIN_FILE_LIST: "./contrib/MechanicalIndustryMeter/mini_mechanical_industry_meter_data/train_mini.txt"
     TRAIN_TOTAL_IMAGES: 64
-    VAL_FILE_LIST: "./dataset/mini_mechanical_industry_meter_data/val_mini.txt"
+    VAL_FILE_LIST: "./contrib/MechanicalIndustryMeter/mini_mechanical_industry_meter_data/val_mini.txt"
     VAL_TOTAL_IMAGES: 8
     SEPARATOR: "|"
     IGNORE_INDEX: 255
diff --git a/contrib/README.md b/contrib/README.md
index c39141255acc41ea4f039b65fae8ca3604c7b972..225ffb7747dba76b3dc3db2b27c764868a5e4fc5 100644
--- a/contrib/README.md
+++ b/contrib/README.md
@@ -1,72 +1,139 @@
 # PaddleSeg 特色垂类分割模型
 
-提供基于PaddlePaddle最新的分割特色模型
+提供基于PaddlePaddle最新的分割特色模型:
 
-## Augmented Context Embedding with Edge Perceiving (ACE2P)
+- [人像分割](#人像分割)
+- [人体解析](#人体解析)
+- [车道线分割](#车道线分割)
+- [工业用表分割](#工业用表分割)
+- [在线体验](#在线体验)
 
+## 人像分割
 
-### 1. 模型概述
-
-CVPR 19 Look into Person (LIP) 单人人像分割比赛冠军模型，详见[ACE2P](./ACE2P)
+**Note:** 本章节所有命令均在`contrib/HumanSeg`目录下执行。
 
-### 2. 模型下载
+```
+cd contrib/HumanSeg
+```
 
-点击[链接](https://paddleseg.bj.bcebos.com/models/ACE2P.tgz)，下载, 在contrib/ACE2P下解压, `tar -xzf ACE2P.tgz`
+### 1. 模型结构
 
-### 3. 数据下载
+DeepLabv3+ backbone为Xception65
 
-前往LIP数据集官网: http://47.100.21.47:9999/overview.php 或点击 [Baidu_Drive](https://pan.baidu.com/s/1nvqmZBN#list/path=%2Fsharelink2787269280-523292635003760%2FLIP%2FLIP&parentPath=%2Fsharelink2787269280-523292635003760), 
+### 2. 下载模型和数据
+ 
+执行以下命令下载并解压模型和数据集：
 
-加载Testing_images.zip, 解压到contrib/ACE2P/data文件夹下
+```
+python download_HumanSeg.py
+```
 
+或点击[链接](https://paddleseg.bj.bcebos.com/models/HumanSeg.tgz)进行手动下载，并解压到contrib/HumanSeg文件夹下
 
-### 4. 运行
 
-**NOTE:** 运行该模型需要2G左右显存
+### 3. 运行
 
-使用GPU预测
+使用GPU预测：
 ```
-python -u infer.py --example ACE2P --use_gpu
+python -u infer.py --example HumanSeg --use_gpu
 ```
 
+
 使用CPU预测：
 ```
-python -u infer.py --example ACE2P
+python -u infer.py --example HumanSeg
 ```
 
-## 人像分割 (HumanSeg)
 
+预测结果存放在contrib/HumanSeg/HumanSeg/result目录下。
 
-### 1. 模型结构
+### 4. 预测结果示例：
 
-DeepLabv3+ backbone为Xception65
+  原图：
+  
+  ![](HumanSeg/imgs/Human.jpg)
+  
+  预测结果：
+  
+  ![](HumanSeg/imgs/HumanSeg.jpg)
 
-### 2. 下载模型和数据
- 
-点击[链接](https://paddleseg.bj.bcebos.com/models/HumanSeg.tgz)，下载解压到contrib文件夹下
 
 
-### 3. 运行
+## 人体解析
+
+![](ACE2P/imgs/result.jpg)
+
+人体解析(Human Parsing)是细粒度的语义分割任务，旨在识别像素级别的人类图像的组成部分（例如，身体部位和服装）。本章节使用冠军模型Augmented Context Embedding with Edge Perceiving (ACE2P)进行预测分割。
+
+
+**Note:** 本章节所有命令均在`contrib/ACE2P`目录下执行。
 
-使用GPU预测：
 ```
-python -u infer.py --example HumanSeg --use_gpu
+cd contrib/ACE2P
 ```
 
+### 1. 模型概述
+
+Augmented Context Embedding with Edge Perceiving (ACE2P)通过融合底层特征、全局上下文信息和边缘细节，端到端训练学习人体解析任务。以ACE2P单人人体解析网络为基础的解决方案在CVPR2019第三届Look into Person (LIP)挑战赛中赢得了全部三个人体解析任务的第一名。详情请参见[ACE2P](./ACE2P)
+
+### 2. 模型下载
+
+执行以下命令下载并解压ACE2P预测模型：
 
-使用CPU预测：
 ```
-python -u infer.py --example HumanSeg
+python download_ACE2P.py
 ```
 
+或点击[链接](https://paddleseg.bj.bcebos.com/models/ACE2P.tgz)进行手动下载, 并在contrib/ACE2P下解压。
 
-### 4. 预测结果示例：
+### 3. 数据下载
+
+测试图片共10000张，
+点击 [Baidu_Drive](https://pan.baidu.com/s/1nvqmZBN#list/path=%2Fsharelink2787269280-523292635003760%2FLIP%2FLIP&parentPath=%2Fsharelink2787269280-523292635003760)
+下载Testing_images.zip，或前往LIP数据集官网进行下载。
+下载后解压到contrib/ACE2P/data文件夹下
 
-  原图：![](imgs/Human.jpg)
+
+### 4. 运行
+
+
+使用GPU预测
+```
+python -u infer.py --example ACE2P --use_gpu
+```
+
+使用CPU预测：
+```
+python -u infer.py --example ACE2P
+```
+
+**NOTE:** 运行该模型需要2G左右显存。由于数据图片较多，预测过程将比较耗时。
+
+#### 5. 预测结果示例：
+  
+  原图：
+  
+  ![](ACE2P/imgs/117676_2149260.jpg)
+  
+  预测结果：
+  
+  ![](ACE2P/imgs/117676_2149260.png)
   
-  预测结果：![](imgs/HumanSeg.jpg)
+### 备注
 
-## 车道线分割 (RoadLine)
+1. 数据及模型路径等详细配置见ACE2P/HumanSeg/RoadLine下的config.py文件
+2. ACE2P模型需预留2G显存，若显存超可调小FLAGS_fraction_of_gpu_memory_to_use
+
+
+
+
+## 车道线分割
+
+**Note:** 本章节所有命令均在`contrib/RoadLine`目录下执行。
+
+```
+cd contrib/RoadLine
+```
 
 ### 1. 模型结构
 
@@ -75,7 +142,15 @@ Deeplabv3+ backbone为MobileNetv2
 
 ### 2. 下载模型和数据
 
-点击[链接](https://paddleseg.bj.bcebos.com/inference_model/RoadLine.tgz)，下载解压在contrib文件夹下
+ 
+执行以下命令下载并解压模型和数据集：
+
+```
+python download_RoadLine.py
+```
+
+或点击[链接](https://paddleseg.bj.bcebos.com/inference_model/RoadLine.tgz)进行手动下载，并解压到contrib/RoadLine文件夹下
+
 
 ### 3. 运行
   
@@ -92,43 +167,75 @@ python -u infer.py --example RoadLine --use_gpu
 python -u infer.py --example RoadLine
 ```
 
+预测结果存放在contrib/RoadLine/RoadLine/result目录下。
 
 #### 4. 预测结果示例：
   
-  原图：![](imgs/RoadLine.jpg)
+  原图：
+  
+  ![](RoadLine/imgs/RoadLine.jpg)
   
-  预测结果：![](imgs/RoadLine.png)
+  预测结果：
+  
+  ![](RoadLine/imgs/RoadLine.png)
+
+
 
 ## 工业用表分割
 
+
+**Note:** 本章节所有命令均在`PaddleSeg`目录下执行。
+
 ### 1. 模型结构
 
 unet
 
 ### 2. 数据准备
  
-cd到PaddleSeg/dataset文件夹下，执行download_mini_mechanical_industry_meter.py
+执行以下命令下载并解压数据集，数据集将存放在contrib/MechanicalIndustryMeter文件夹下：
+
+```
+python ./contrib/MechanicalIndustryMeter/download_mini_mechanical_industry_meter.py
+```
 
 
-### 3. 训练与评估
+### 3. 下载预训练模型
 
 ```
-CUDA_VISIBLE_DEVICES=0 python ./pdseg/train.py --log_steps 10 --cfg configs/unet_mechanical_meter.yaml --use_gpu --do_eval --use_mpio 
+python ./pretrained_model/download_model.py unet_bn_coco
 ```
 
-### 4. 可视化
-我们提供了一个训练好的模型，点击[链接](https://paddleseg.bj.bcebos.com/models/unet_mechanical_industry_meter.tar)，下载后放在PaddleSeg/pretrained_model下
+### 4. 训练与评估
+
 ```
-CUDA_VISIBLE_DEVICES=0 python ./pdseg/vis.py --cfg configs/unet_mechanical_meter.yaml --use_gpu --vis_dir vis_meter \
-TEST.TEST_MODEL "./pretrained_model/unet_gongyeyongbiao/" 
+export CUDA_VISIBLE_DEVICES=0 
+python ./pdseg/train.py --log_steps 10 --cfg contrib/MechanicalIndustryMeter/unet_mechanical_meter.yaml --use_gpu --do_eval --use_mpio 
 ```
-可视化结果会保存在vis_meter文件夹下
 
-### 5. 可视化结果示例：
+### 5. 可视化
+我们已提供了一个训练好的模型，执行以下命令进行下载，下载后将存放在./contrib/MechanicalIndustryMeter/文件夹下。
 
-  原图：![](imgs/1560143028.5_IMG_3091.JPG)
+```
+python ./contrib/MechanicalIndustryMeter/download_unet_mechanical_industry_meter.py
+```
+
+使用该模型进行预测可视化：
+
+```
+python ./pdseg/vis.py --cfg contrib/MechanicalIndustryMeter/unet_mechanical_meter.yaml --use_gpu --vis_dir vis_meter \
+TEST.TEST_MODEL "./contrib/MechanicalIndustryMeter/unet_mechanical_industry_meter/" 
+```
+可视化结果会保存在./vis_meter文件夹下。
+
+### 6. 可视化结果示例：
+
+  原图：
+  
+  ![](MechanicalIndustryMeter/imgs/1560143028.5_IMG_3091.JPG)
   
-  预测结果：![](imgs/1560143028.5_IMG_3091.png)
+  预测结果：
+  
+  ![](MechanicalIndustryMeter/imgs/1560143028.5_IMG_3091.png)
 
 ## 在线体验
 
@@ -140,7 +247,4 @@ PaddleSeg在AI Studio平台上提供了在线体验的教程，欢迎体验：
 |人像分割|[点击体验](https://aistudio.baidu.com/aistudio/projectdetail/188833)|
 |特色垂类模型|[点击体验](https://aistudio.baidu.com/aistudio/projectdetail/115541)|
 
-## 备注
 
-1. 数据及模型路径等详细配置见ACE2P/HumanSeg/RoadLine下的config.py文件
-2. ACE2P模型需预留2G显存，若显存超可调小FLAGS_fraction_of_gpu_memory_to_use
diff --git a/contrib/RoadLine/download_RoadLine.py b/contrib/RoadLine/download_RoadLine.py
new file mode 100644
index 0000000000000000000000000000000000000000..86b631784edadcff6d575c59e67ee23a1775216d
--- /dev/null
+++ b/contrib/RoadLine/download_RoadLine.py
@@ -0,0 +1,31 @@
+# Copyright (c) 2019  PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import sys
+import os
+
+LOCAL_PATH = os.path.dirname(os.path.abspath(__file__))
+TEST_PATH = os.path.join(LOCAL_PATH, "..", "..", "test")
+sys.path.append(TEST_PATH)
+
+from test_utils import download_file_and_uncompress
+
+if __name__ == "__main__":
+    download_file_and_uncompress(
+        url='https://paddleseg.bj.bcebos.com/inference_model/RoadLine.tgz',
+        savepath=LOCAL_PATH,
+        extrapath=LOCAL_PATH,
+        extraname='RoadLine')
+
+    print("Pretrained Model download success!")
diff --git a/contrib/imgs/RoadLine.jpg b/contrib/RoadLine/imgs/RoadLine.jpg
similarity index 100%
rename from contrib/imgs/RoadLine.jpg
rename to contrib/RoadLine/imgs/RoadLine.jpg
diff --git a/contrib/imgs/RoadLine.png b/contrib/RoadLine/imgs/RoadLine.png
similarity index 100%
rename from contrib/imgs/RoadLine.png
rename to contrib/RoadLine/imgs/RoadLine.png
diff --git a/contrib/RoadLine/infer.py b/contrib/RoadLine/infer.py
new file mode 100644
index 0000000000000000000000000000000000000000..971476933c431977ce80c73e1d939fe079e1af19
--- /dev/null
+++ b/contrib/RoadLine/infer.py
@@ -0,0 +1,130 @@
+# -*- coding: utf-8 -*-
+import os
+import cv2
+import numpy as np
+from utils.util import get_arguments
+from utils.palette import get_palette
+from PIL import Image as PILImage
+import importlib
+
+args = get_arguments()
+config = importlib.import_module('config')
+cfg = getattr(config, 'cfg')
+
+# paddle垃圾回收策略FLAG，ACE2P模型较大，当显存不够时建议开启
+os.environ['FLAGS_eager_delete_tensor_gb']='0.0'
+
+import paddle.fluid as fluid
+
+# 预测数据集类
+class TestDataSet():
+    def __init__(self):
+        self.data_dir = cfg.data_dir 
+        self.data_list_file = cfg.data_list_file
+        self.data_list = self.get_data_list()
+        self.data_num = len(self.data_list)
+    
+    def get_data_list(self):
+        # 获取预测图像路径列表
+        data_list = []
+        data_file_handler = open(self.data_list_file, 'r')
+        for line in data_file_handler:
+            img_name = line.strip()
+            name_prefix = img_name.split('.')[0]
+            if len(img_name.split('.')) == 1:
+                img_name = img_name + '.jpg'
+            img_path = os.path.join(self.data_dir, img_name)
+            data_list.append(img_path)
+        return data_list
+
+    def preprocess(self, img):
+        # 图像预处理
+        if cfg.example == 'ACE2P':
+            reader = importlib.import_module(args.example+'.reader')
+            ACE2P_preprocess = getattr(reader, 'preprocess')
+            img = ACE2P_preprocess(img)
+        else:
+            img = cv2.resize(img, cfg.input_size).astype(np.float32)
+            img -= np.array(cfg.MEAN)
+            img /= np.array(cfg.STD)
+            img = img.transpose((2, 0, 1))
+            img = np.expand_dims(img, axis=0)
+        return img
+
+    def get_data(self, index):
+        # 获取图像信息
+        img_path = self.data_list[index]
+        img = cv2.imread(img_path, cv2.IMREAD_COLOR)
+        if img is None:
+            return img, img,img_path, None
+
+        img_name = img_path.split(os.sep)[-1]
+        name_prefix = img_name.replace('.'+img_name.split('.')[-1],'')
+        img_shape = img.shape[:2]
+        img_process = self.preprocess(img)
+
+        return img, img_process, name_prefix, img_shape
+
+
+def infer():
+    if not os.path.exists(cfg.vis_dir):
+        os.makedirs(cfg.vis_dir)
+    palette = get_palette(cfg.class_num)
+    # 人像分割结果显示阈值
+    thresh = 120
+
+    place = fluid.CUDAPlace(0) if cfg.use_gpu else fluid.CPUPlace()
+    exe = fluid.Executor(place)
+
+    # 加载预测模型
+    test_prog, feed_name, fetch_list = fluid.io.load_inference_model(
+        dirname=cfg.model_path, executor=exe, params_filename='__params__')
+
+    #加载预测数据集
+    test_dataset = TestDataSet()
+    data_num = test_dataset.data_num
+
+    for idx in range(data_num):
+        # 数据获取
+        ori_img, image, im_name, im_shape = test_dataset.get_data(idx)
+        if image is None:
+            print(im_name, 'is None')
+            continue
+        
+        # 预测
+        if cfg.example == 'ACE2P':
+            # ACE2P模型使用多尺度预测
+            reader = importlib.import_module(args.example+'.reader')
+            multi_scale_test = getattr(reader, 'multi_scale_test')
+            parsing, logits = multi_scale_test(exe, test_prog, feed_name, fetch_list, image, im_shape)
+        else:
+            # HumanSeg,RoadLine模型单尺度预测
+            result = exe.run(program=test_prog, feed={feed_name[0]: image}, fetch_list=fetch_list)
+            parsing = np.argmax(result[0][0], axis=0)
+            parsing = cv2.resize(parsing.astype(np.uint8), im_shape[::-1])
+        
+        # 预测结果保存
+        result_path = os.path.join(cfg.vis_dir, im_name + '.png')
+        if cfg.example == 'HumanSeg':
+            logits = result[0][0][1]*255
+            logits = cv2.resize(logits, im_shape[::-1])
+            ret, logits = cv2.threshold(logits, thresh, 0, cv2.THRESH_TOZERO)
+            logits = 255 *(logits - thresh)/(255 - thresh)
+            # 将分割结果添加到alpha通道
+            rgba = np.concatenate((ori_img, np.expand_dims(logits, axis=2)), axis=2)
+            cv2.imwrite(result_path, rgba)
+        else: 
+            output_im = PILImage.fromarray(np.asarray(parsing, dtype=np.uint8))
+            output_im.putpalette(palette)
+            output_im.save(result_path)
+
+        if (idx + 1) % 100 == 0:
+            print('%d  processd' % (idx + 1))
+            
+    print('%d  processd done' % (idx + 1))   
+    
+    return 0
+
+
+if __name__ == "__main__":
+    infer()
diff --git a/contrib/RoadLine/utils/__init__.py b/contrib/RoadLine/utils/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/contrib/RoadLine/utils/palette.py b/contrib/RoadLine/utils/palette.py
new file mode 100644
index 0000000000000000000000000000000000000000..2186203cbc2789f6eff70dfd92f724b4fe16cdb7
--- /dev/null
+++ b/contrib/RoadLine/utils/palette.py
@@ -0,0 +1,38 @@
+##+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+## Created by: RainbowSecret
+## Microsoft Research
+## yuyua@microsoft.com
+## Copyright (c) 2018
+##
+## This source code is licensed under the MIT-style license found in the
+## LICENSE file in the root directory of this source tree
+##+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+import numpy as np
+import cv2
+
+
+def get_palette(num_cls):
+    """ Returns the color map for visualizing the segmentation mask.
+    Args:
+        num_cls: Number of classes
+    Returns:
+        The color map
+    """
+    n = num_cls
+    palette = [0] * (n * 3)
+    for j in range(0, n):
+        lab = j
+        palette[j * 3 + 0] = 0
+        palette[j * 3 + 1] = 0
+        palette[j * 3 + 2] = 0
+        i = 0
+        while lab:
+            palette[j * 3 + 0] |= (((lab >> 0) & 1) << (7 - i))
+            palette[j * 3 + 1] |= (((lab >> 1) & 1) << (7 - i))
+            palette[j * 3 + 2] |= (((lab >> 2) & 1) << (7 - i))
+            i += 1
+            lab >>= 3
+    return palette
diff --git a/contrib/RoadLine/utils/util.py b/contrib/RoadLine/utils/util.py
new file mode 100644
index 0000000000000000000000000000000000000000..7394870e7c94c1fb16169e314696b931eecdc3b2
--- /dev/null
+++ b/contrib/RoadLine/utils/util.py
@@ -0,0 +1,47 @@
+from __future__ import division
+from __future__ import print_function
+from __future__ import unicode_literals
+import argparse
+import os
+
+def get_arguments():
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--use_gpu",
+                        action="store_true",
+                        help="Use gpu or cpu to test.")
+    parser.add_argument('--example',
+                        type=str,
+                        help='RoadLine, HumanSeg or ACE2P')
+                        
+    return parser.parse_args()
+
+
+class AttrDict(dict):
+    def __init__(self, *args, **kwargs):
+        super(AttrDict, self).__init__(*args, **kwargs)
+
+    def __getattr__(self, name):
+        if name in self.__dict__:
+            return self.__dict__[name]
+        elif name in self:
+            return self[name]
+        else:
+            raise AttributeError(name)
+
+    def __setattr__(self, name, value):
+        if name in self.__dict__:
+            self.__dict__[name] = value
+        else:
+            self[name] = value
+
+def merge_cfg_from_args(args, cfg):
+    """Merge config keys, values in args into the global config."""
+    for k, v in vars(args).items():
+        d = cfg
+        try:
+            value = eval(v)
+        except:
+            value = v
+        if value is not None:
+            cfg[k] = value
+