document polish (#149)

* update all docs involving color label

configs/deeplabv3p_xception65_cityscapes.yaml

+EVAL_CROP_SIZE: (2049, 1025) # (width, height), for unpadding rangescaling and stepscaling
+TRAIN_CROP_SIZE: (769, 769) # (width, height), for unpadding rangescaling and stepscaling
+AUG:
+    AUG_METHOD: "stepscaling" # choice unpadding rangescaling and stepscaling
+    FIX_RESIZE_SIZE: (2048, 1024) # (width, height), for unpadding
+    INF_RESIZE_VALUE: 500  # for rangescaling
+    MAX_RESIZE_VALUE: 600  # for rangescaling
+    MIN_RESIZE_VALUE: 400  # for rangescaling
+    MAX_SCALE_FACTOR: 2.0  # for stepscaling
+    MIN_SCALE_FACTOR: 0.5  # for stepscaling
+    SCALE_STEP_SIZE: 0.25  # for stepscaling
+    MIRROR: True
+BATCH_SIZE: 4
+DATASET:
+    DATA_DIR: "./dataset/cityscapes/"
+    IMAGE_TYPE: "rgb"  # choice rgb or rgba
+    NUM_CLASSES: 19
+    TEST_FILE_LIST: "dataset/cityscapes/val.list"
+    TRAIN_FILE_LIST: "dataset/cityscapes/train.list"
+    VAL_FILE_LIST: "dataset/cityscapes/val.list"
+    IGNORE_INDEX: 255
+    SEPARATOR: " "
+FREEZE:
+    MODEL_FILENAME: "model"
+    PARAMS_FILENAME: "params"
+MODEL:
+    DEFAULT_NORM_TYPE: "bn"
+    MODEL_NAME: "deeplabv3p"
+    DEEPLAB:
+        ASPP_WITH_SEP_CONV: True
+        DECODER_USE_SEP_CONV: True
+TRAIN:
+    PRETRAINED_MODEL_DIR: u"pretrained_model/deeplabv3p_xception65_bn_coco"
+    MODEL_SAVE_DIR: "saved_model/deeplabv3p_xception65_bn_cityscapes"
+    SNAPSHOT_EPOCH: 10
+    SYNC_BATCH_NORM: True
+TEST:
+    TEST_MODEL: "saved_model/deeplabv3p_xception65_bn_cityscapes/final"
+SOLVER:
+    LR: 0.01
+    LR_POLICY: "poly"
+    OPTIMIZER: "sgd"
+    NUM_EPOCHS: 100

docs/data_prepare.md

@@ -19,7 +19,7 @@ PaddleSeg采用单通道的标注图片，每一种像素值代表一种类别
 
 ### 灰度标注转换为伪彩色标注
 如果用户需要转换成伪彩色标注图，可使用我们的转换工具。适用于以下两种常见的情况：
-1. 从指定灰度标注所在的目录读取标注图片
+1. 如果您希望将指定目录下的所有灰度标注图转换为伪彩色标注图，则执行以下命令，指定灰度标注所在的目录即可。
 ```buildoutcfg
 python pdseg/tools/gray2pseudo_color.py <dir_or_file> <output_dir>
 ```
@@ -29,7 +29,7 @@ python pdseg/tools/gray2pseudo_color.py <dir_or_file> <output_dir>
 |dir_or_file|指定灰度标注所在目录|
 |output_dir|彩色标注图片的输出目录|
 
-2. 从已有文件列表中读取标注图片
+2. 如果您仅希望将指定数据集中的部分灰度标注图转换为伪彩色标注图，则执行以下命令，需要已有文件列表，按列表读取指定图片。
 ```buildoutcfg
 python pdseg/tools/gray2pseudo_color.py <dir_or_file> <output_dir> --dataset_dir <dataset directory> --file_separator <file list separator>
 ```

docs/model_zoo.md

 # PaddleSeg 预训练模型
 
-PaddleSeg对所有内置的分割模型都提供了公开数据集下的预训练模型，通过加载预训练模型后训练可以在自定义数据集中得到更稳定地效果。
+PaddleSeg对所有内置的分割模型都提供了公开数据集下的预训练模型。因为对于自定
+义数据集的场景，使用预训练模型进行训练可以得到更稳定地效果。用户可以根据模型类型、自己的数据集和预训练数据集的相似程度，选择并下载预训练模型。
 
 ## ImageNet预训练模型
 
@@ -32,6 +33,11 @@ PaddleSeg对所有内置的分割模型都提供了公开数据集下的预训
 | HRNet_W48 | ImageNet | [hrnet_w48_imagenet.tar](https://paddleseg.bj.bcebos.com/models/hrnet_w48_imagenet.tar) | 78.95%/94.42% |
 | HRNet_W64 | ImageNet | [hrnet_w64_imagenet.tar](https://paddleseg.bj.bcebos.com/models/hrnet_w64_imagenet.tar) | 79.30%/94.61% |
 
+| 模型 | 数据集合 | 下载地址 | Accuray Top1/5 Error |
+|---|---|---|---|
+| ResNet50（适配PSPNet） | ImageNet | [resnet50_v2_pspnet](https://paddleseg.bj.bcebos.com/resnet50_v2_pspnet.tgz)| -- |
+| ResNet101（适配PSPNet） | ImageNet | [resnet101_v2_pspnet](https://paddleseg.bj.bcebos.com/resnet101_v2_pspnet.tgz)| -- |
+
 ## COCO预训练模型
 
 数据集为COCO实例分割数据集合转换成的语义分割数据集合

turtorial/finetune_pspnet.md

@@ -132,7 +132,9 @@ python pdseg/vis.py --use_gpu --cfg ./configs/pspnet_optic.yaml
 
 |模型|BackBone|数据集|配置|
 |-|-|-|-|
-|[pspnet50_cityscapes](https://paddleseg.bj.bcebos.com/models/pspnet50_cityscapes.tgz)|ResNet50|Cityscapes |MODEL.MODEL_NAME: pspnet <br> MODEL.DEFAULT_NORM_TYPE: bn <br> MODEL.PSPNET.LAYERS: 50|
-|[pspnet101_cityscapes](https://paddleseg.bj.bcebos.com/models/pspnet101_cityscapes.tgz)|ResNet101|Cityscapes |MODEL.MODEL_NAME: pspnet <br> MODEL.DEFAULT_NORM_TYPE: bn <br> MODEL.PSPNET.LAYERS: 101|
-| [pspnet50_coco](https://paddleseg.bj.bcebos.com/models/pspnet50_coco.tgz)|ResNet50|COCO  |MODEL.MODEL_NAME: pspnet <br> MODEL.DEFAULT_NORM_TYPE: bn <br> MODEL.PSPNET.LAYERS: 50|
-| [pspnet101_coco](https://paddleseg.bj.bcebos.com/models/pspnet101_coco.tgz) |ResNet101| COCO |MODEL.MODEL_NAME: pspnet <br> MODEL.DEFAULT_NORM_TYPE: bn <br> MODEL.PSPNET.LAYERS: 101|
+|[pspnet50_cityscapes](https://paddleseg.bj.bcebos.com/models/pspnet50_cityscapes.tgz)|ResNet50（适配PSPNet）|Cityscapes |MODEL.MODEL_NAME: pspnet <br> MODEL.DEFAULT_NORM_TYPE: bn <br> MODEL.PSPNET.LAYERS: 50|
+|[pspnet101_cityscapes](https://paddleseg.bj.bcebos.com/models/pspnet101_cityscapes.tgz)|ResNet101（适配PSPNet）|Cityscapes |MODEL.MODEL_NAME: pspnet <br> MODEL.DEFAULT_NORM_TYPE: bn <br> MODEL.PSPNET.LAYERS: 101|
+| [pspnet50_coco](https://paddleseg.bj.bcebos.com/models/pspnet50_coco.tgz)|ResNet50（适配PSPNet）|COCO  |MODEL.MODEL_NAME: pspnet <br> MODEL.DEFAULT_NORM_TYPE: bn <br> MODEL.PSPNET.LAYERS: 50|
+| [pspnet101_coco](https://paddleseg.bj.bcebos.com/models/pspnet101_coco.tgz) |ResNet101（适配PSPNet）| COCO |MODEL.MODEL_NAME: pspnet <br> MODEL.DEFAULT_NORM_TYPE: bn <br> MODEL.PSPNET.LAYERS: 101|
+| [resnet50_v2_pspnet](https://paddleseg.bj.bcebos.com/resnet50_v2_pspnet.tgz)| ResNet50（适配PSPNet） | ImageNet | MODEL.MODEL_NAME: pspnet <br> MODEL.DEFAULT_NORM_TYPE: bn <br> MODEL.PSPNET.LAYERS: 50 |
+| [resnet101_v2_pspnet](https://paddleseg.bj.bcebos.com/resnet101_v2_pspnet.tgz)| ResNet101（适配PSPNet） | ImageNet | MODEL.MODEL_NAME: pspnet <br> MODEL.DEFAULT_NORM_TYPE: bn <br> MODEL.PSPNET.LAYERS: 101 |