optimize environment configuration (#521)

* optimize python path

dataset/coco/download_coco.py

@@ -15,6 +15,10 @@
 import sys
 import os.path as osp
 import logging
+# add python path of PadleDetection to sys.path
+parent_path = osp.abspath(osp.join(__file__, *(['..'] * 3)))
+if parent_path not in sys.path:
+    sys.path.append(parent_path)
 
 from ppdet.utils.download import download_dataset
 

dataset/fruit/download_fruit.py

@@ -15,6 +15,10 @@
 import sys
 import os.path as osp
 import logging
+# add python path of PadleDetection to sys.path
+parent_path = osp.abspath(osp.join(__file__, *(['..'] * 3)))
+if parent_path not in sys.path:
+    sys.path.append(parent_path)
 
 from ppdet.utils.download import download_dataset
 

dataset/voc/create_list.py

@@ -15,6 +15,10 @@
 import sys
 import os.path as osp
 import logging
+# add python path of PadleDetection to sys.path
+parent_path = osp.abspath(osp.join(__file__, *(['..'] * 3)))
+if parent_path not in sys.path:
+    sys.path.append(parent_path)
 
 from ppdet.utils.download import create_voc_list
 

dataset/voc/download_voc.py

@@ -15,6 +15,10 @@
 import sys
 import os.path as osp
 import logging
+# add python path of PadleDetection to sys.path
+parent_path = osp.abspath(osp.join(__file__, *(['..'] * 3)))
+if parent_path not in sys.path:
+    sys.path.append(parent_path)
 
 from ppdet.utils.download import download_dataset
 

docs/advanced_tutorials/TRANSFER_LEARNING.md

@@ -15,7 +15,6 @@ In transfer learning, it's needed to load pretrained model selectively. Two ways
 The parameters which have diffierent shape between model and pretrain\_weights are ignored automatically. For example:
 
 ```python
-export PYTHONPATH=$PYTHONPATH:.
 export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
 python -u tools/train.py -c configs/faster_rcnn_r50_1x.yml \
                       -o pretrain_weights=https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_r50_1x.tar
@@ -29,7 +28,6 @@ The parameters which need to ignore can be specified explicitly as well and arbi
 - Set `finetune_exclude_pretrained_params` in command line. For example:
 
 ```python
-export PYTHONPATH=$PYTHONPATH:.
 export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
 python -u tools/train.py -c configs/faster_rcnn_r50_1x.yml \
                         -o pretrain_weights=https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_r50_1x.tar \

docs/advanced_tutorials/TRANSFER_LEARNING_cn.md

@@ -13,7 +13,6 @@
 模型中和预训练模型中对应参数形状不同的参数将自动被忽略，例如：
 
 ```python
-export PYTHONPATH=$PYTHONPATH:.
 export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
 python -u tools/train.py -c configs/faster_rcnn_r50_1x.yml \
                            -o pretrain_weights=https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_r50_1x.tar
@@ -28,7 +27,6 @@ python -u tools/train.py -c configs/faster_rcnn_r50_1x.yml \
 2. 在 train.py的启动参数中设置`finetune_exclude_pretrained_params`。例如：
 
 ```python
-export PYTHONPATH=$PYTHONPATH:.
 export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
 python -u tools/train.py -c configs/faster_rcnn_r50_1x.yml \
                          -o pretrain_weights=https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_r50_1x.tar \

docs/advanced_tutorials/inference/EXPORT_MODEL.md

@@ -4,7 +4,7 @@
 
 **说明：**
 
-- **输入部分：**导出模型输入为网络输入图像，即原始图片经过预处理后的图像，具体预处理方式可参考配置文件中TestReader部分。各类检测模型的输入格式分别为：
+- **输入部分：** 导出模型输入为网络输入图像，即原始图片经过预处理后的图像，具体预处理方式可参考配置文件中TestReader部分。各类检测模型的输入格式分别为：
 
 | 模型系列名称 | 输入图像预处理方式 | 其他输入信息 |
 | :---------: | ----------- | ---------- |
@@ -16,7 +16,7 @@
 | Face   |  归一化  | im\_shape: 格式为[origin\_H, origin\_W], origin为原始图像  |
 
 
-- **输出部分：**导出模型输出统一为NMS的输出，形状为[N, 6], 其中N为预测框的个数，6为[class_id, score, x1, y1, x2, y2]。
+- **输出部分：** 导出模型输出统一为NMS的输出，形状为[N, 6], 其中N为预测框的个数，6为[class_id, score, x1, y1, x2, y2]。
 
 - 模型导出不支持模型结构中包含```fluid.layers.py_func```的情况。
 

docs/featured_model/CONTRIB.md

@@ -43,7 +43,6 @@ Users can employ the model to conduct the inference:
 
 ```
 export CUDA_VISIBLE_DEVICES=0
-export PYTHONPATH=$PYTHONPATH:.
 python -u tools/infer.py -c contrib/VehicleDetection/vehicle_yolov3_darknet.yml \
                          -o weights=https://paddlemodels.bj.bcebos.com/object_detection/vehicle_yolov3_darknet.tar \
                          --infer_dir contrib/VehicleDetection/demo \
@@ -90,7 +89,6 @@ Users can employ the model to conduct the inference:
 
 ```
 export CUDA_VISIBLE_DEVICES=0
-export PYTHONPATH=$PYTHONPATH:.
 python -u tools/infer.py -c contrib/PedestrianDetection/pedestrian_yolov3_darknet.yml \
                          -o weights=https://paddlemodels.bj.bcebos.com/object_detection/pedestrian_yolov3_darknet.tar \
                          --infer_dir contrib/PedestrianDetection/demo \

docs/featured_model/CONTRIB_cn.md

@@ -44,7 +44,6 @@ IOU=.5时的AP为 0.764。
 
 ```
 export CUDA_VISIBLE_DEVICES=0
-export PYTHONPATH=$PYTHONPATH:.
 python -u tools/infer.py -c contrib/VehicleDetection/vehicle_yolov3_darknet.yml \
                          -o weights=https://paddlemodels.bj.bcebos.com/object_detection/vehicle_yolov3_darknet.tar \
                          --infer_dir contrib/VehicleDetection/demo \
@@ -92,7 +91,6 @@ IOU=.5-.95时的AP为 0.518。
 
 ```
 export CUDA_VISIBLE_DEVICES=0
-export PYTHONPATH=$PYTHONPATH:.
 python -u tools/infer.py -c contrib/PedestrianDetection/pedestrian_yolov3_darknet.yml \
                          -o weights=https://paddlemodels.bj.bcebos.com/object_detection/pedestrian_yolov3_darknet.tar \
                          --infer_dir contrib/PedestrianDetection/demo \

docs/featured_model/FACE_DETECTION.md

@@ -155,7 +155,6 @@ cd dataset/wider_face && ./download.sh
 评估并生成结果文件：
 ```
 export CUDA_VISIBLE_DEVICES=0
-export PYTHONPATH=$PYTHONPATH:.
 python -u tools/face_eval.py -c configs/face_detection/blazeface.yml \
        -o weights=output/blazeface/model_final \
        --eval_mode=widerface

docs/featured_model/FACE_DETECTION_en.md

@@ -170,7 +170,6 @@ Default is `False`, it will select `single-scale` evaluation.
 - Evaluate and generate results files:
 ```
 export CUDA_VISIBLE_DEVICES=0
-export PYTHONPATH=$PYTHONPATH:.
 python -u tools/face_eval.py -c configs/face_detection/blazeface.yml \
        -o weights=output/blazeface/model_final \
        --eval_mode=widerface

docs/tutorials/GETTING_STARTED.md

@@ -11,8 +11,6 @@ instructions](INSTALL.md).
 PaddleDetection provides scripots for training, evalution and inference with various features according to different configure.
 
 ```bash
-# set PYTHONPATH
-export PYTHONPATH=$PYTHONPATH:.
 # training in single-GPU and multi-GPU. specify different GPU numbers by CUDA_VISIBLE_DEVICES
 export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
 python tools/train.py -c configs/faster_rcnn_r50_1x.yml

docs/tutorials/GETTING_STARTED_cn.md

@@ -8,8 +8,6 @@
 PaddleDetection提供了训练/评估/推断三个功能的使用脚本，支持通过不同可选参数实现特定功能
 
 ```bash
-# 设置PYTHONPATH路径
-export PYTHONPATH=$PYTHONPATH:.
 # GPU训练 支持单卡，多卡训练，通过CUDA_VISIBLE_DEVICES指定卡号
 export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
 python tools/train.py -c configs/faster_rcnn_r50_1x.yml

docs/tutorials/INSTALL.md

@@ -88,15 +88,6 @@ Required python packages are specified in [requirements.txt](https://github.com/
 pip install -r requirements.txt
 ```
 
-**Specify the current Python path:**
-
-```shell
-# In Linux/Mac
-export PYTHONPATH=$PYTHONPATH:.
-# In windows
-set PYTHONPATH=%PYTHONPATH%;.
-```
-
 **Make sure the tests pass:**
 
 ```shell
@@ -121,7 +112,6 @@ ln -sf <path/to/voc> <path/to/paddle_detection>/dataset/voc
 For Pascal VOC dataset, you should create file list by:
 
 ```
-export PYTHONPATH=$PYTHONPATH:.
 python dataset/voc/create_list.py
 ```
 
@@ -132,7 +122,6 @@ On the other hand, to download the datasets, run the following commands:
 - COCO
 
 ```
-export PYTHONPATH=$PYTHONPATH:.
 python dataset/coco/download_coco.py
 ```
 
@@ -160,7 +149,6 @@ python dataset/coco/download_coco.py
 - Pascal VOC
 
 ```
-export PYTHONPATH=$PYTHONPATH:.
 python dataset/voc/download_voc.py
 python dataset/voc/create_list.py
 ```

docs/tutorials/INSTALL_cn.md

@@ -84,15 +84,6 @@ Python依赖库在[requirements.txt](https://github.com/PaddlePaddle/PaddleDetec
 pip install -r requirements.txt
 ```
 
-**指定当前Python路径：**
-
-```shell
-# 在Linux/Mac系统下运行:
-export PYTHONPATH=$PYTHONPATH:.
-# 在windows系统下运行:
-set PYTHONPATH=%PYTHONPATH%;.
-```
-
 **确认测试通过：**
 
 ```
@@ -119,7 +110,6 @@ ln -sf <path/to/voc> <path/to/paddle_detection>/dataset/voc
 对于Pascal VOC数据集，需通过如下命令创建文件列表：
 
 ```
-export PYTHONPATH=$PYTHONPATH:.
 python dataset/voc/create_list.py
 ```
 
@@ -130,7 +120,6 @@ python dataset/voc/create_list.py
 - COCO
 
 ```
-export PYTHONPATH=$PYTHONPATH:.
 python dataset/coco/download_coco.py
 ```
 
@@ -158,7 +147,6 @@ python dataset/coco/download_coco.py
 - Pascal VOC
 
 ```
-export PYTHONPATH=$PYTHONPATH:.
 python dataset/voc/download_voc.py
 python dataset/voc/create_list.py
 ```

docs/tutorials/QUICK_STARTED.md

@@ -4,10 +4,9 @@ English | [简体中文](QUICK_STARTED_cn.md)
 
 This tutorial fine-tunes a tiny dataset by pretrained detection model for users to get a model and learn PaddleDetection quickly. The model can be trained in around 20min with good performance.
 
-- **Note: before started, need to set PYTHONPATH and specifiy the GPU device as follows in Linux platform. For Windows users, also need to set PYTHONPATH correctly.**
+- **Note: before started, need to specifiy the GPU device as follows.**
 
 ```bash
-export PYTHONPATH=$PYTHONPATH:.
 export CUDA_VISIBLE_DEVICES=0
 ```
 

docs/tutorials/QUICK_STARTED_cn.md

@@ -4,10 +4,9 @@
 
 为了使得用户能够在很短的时间内快速产出模型，掌握PaddleDetection的使用方式，这篇教程通过一个预训练检测模型对小数据集进行finetune。在P40上单卡大约20min即可产出一个效果不错的模型。
 
-- **注：在开始前，对于Linux用户, 运行如下命令设置PYTHONPATH，如果有GPU设备，指定GPU设备号。对于Windows用户，同样需要将PaddleDetection的根目录设置到PYTHONPATH中。**
+- **注：在开始前，如果有GPU设备，指定GPU设备号。**
 
 ```bash
-export PYTHONPATH=$PYTHONPATH:.
 export CUDA_VISIBLE_DEVICES=0
 ```
 

ppdet/data/tests/test_dataset.py

@@ -18,6 +18,10 @@ import sys
 import logging
 import random
 import copy
+# add python path of PadleDetection to sys.path
+parent_path = os.path.abspath(os.path.join(__file__, *(['..'] * 4)))
+if parent_path not in sys.path:
+    sys.path.append(parent_path)
 
 from ppdet.data.parallel_map import ParallelMap
 

ppdet/data/tests/test_loader.py

@@ -14,6 +14,11 @@
 
 import unittest
 import os
+import sys
+# add python path of PadleDetection to sys.path
+parent_path = os.path.abspath(os.path.join(__file__, *(['..'] * 4)))
+if parent_path not in sys.path:
+    sys.path.append(parent_path)
 
 from ppdet.data.source.coco import COCODataSet
 from ppdet.data.reader import Reader

ppdet/data/tests/test_loader_yaml.py

@@ -16,6 +16,11 @@ import unittest
 import os
 import yaml
 import logging
+import sys
+# add python path of PadleDetection to sys.path
+parent_path = os.path.abspath(os.path.join(__file__, *(['..'] * 4)))
+if parent_path not in sys.path:
+    sys.path.append(parent_path)
 
 from ppdet.utils.download import get_path
 from ppdet.utils.download import DATASET_HOME

ppdet/modeling/tests/test_architectures.py

@@ -20,6 +20,12 @@ import unittest
 import numpy as np
 
 import paddle.fluid as fluid
+import os
+import sys
+# add python path of PadleDetection to sys.path
+parent_path = os.path.abspath(os.path.join(__file__, *(['..'] * 4)))
+if parent_path not in sys.path:
+    sys.path.append(parent_path)
 
 from ppdet.modeling.tests.decorator_helper import prog_scope
 from ppdet.core.workspace import load_config, merge_config, create

tools/configure.py

@@ -14,8 +14,14 @@
 
 from __future__ import print_function
 
+import os
 import sys
 from argparse import ArgumentParser, RawDescriptionHelpFormatter
+import sys
+# add python path of PadleDetection to sys.path
+parent_path = os.path.abspath(os.path.join(__file__, *(['..'] * 2)))
+if parent_path not in sys.path:
+    sys.path.append(parent_path)
 
 import yaml
 

tools/eval.py

@@ -17,6 +17,11 @@ from __future__ import division
 from __future__ import print_function
 
 import os
+import sys
+# add python path of PadleDetection to sys.path
+parent_path = os.path.abspath(os.path.join(__file__, *(['..'] * 2)))
+if parent_path not in sys.path:
+    sys.path.append(parent_path)
 
 
 def set_paddle_flags(**kwargs):

tools/export_model.py

@@ -17,6 +17,11 @@ from __future__ import division
 from __future__ import print_function
 
 import os
+import sys
+# add python path of PadleDetection to sys.path
+parent_path = os.path.abspath(os.path.join(__file__, *(['..'] * 2)))
+if parent_path not in sys.path:
+    sys.path.append(parent_path)
 
 from paddle import fluid
 

tools/face_eval.py

@@ -17,6 +17,11 @@ from __future__ import division
 from __future__ import print_function
 
 import os
+import sys
+# add python path of PadleDetection to sys.path
+parent_path = os.path.abspath(os.path.join(__file__, *(['..'] * 2)))
+if parent_path not in sys.path:
+    sys.path.append(parent_path)
 
 import paddle.fluid as fluid
 import numpy as np

tools/infer.py

@@ -18,6 +18,11 @@ from __future__ import print_function
 
 import os
 import glob
+import sys
+# add python path of PadleDetection to sys.path
+parent_path = os.path.abspath(os.path.join(__file__, *(['..'] * 2)))
+if parent_path not in sys.path:
+    sys.path.append(parent_path)
 
 import numpy as np
 from PIL import Image

tools/train.py

@@ -23,6 +23,11 @@ import random
 import datetime
 from collections import deque
 from paddle.fluid import profiler
+import sys
+# add python path of PadleDetection to sys.path
+parent_path = os.path.abspath(os.path.join(__file__, *(['..'] * 2)))
+if parent_path not in sys.path:
+    sys.path.append(parent_path)
 
 
 def set_paddle_flags(**kwargs):