Update SSD documentation. #968

sefira · 2018-06-07T07:37:10Z

See https://github.com/sefira/models/tree/ssd_wiki/fluid/object_detection

qingqing01

@sefira Thanks very much!

qingqing01 · 2018-06-11T06:15:29Z

fluid/object_detection/README.md

+<p align="center">
+<img src="images/SSD_paper_figure.jpg" height=300 width=900 hspace='10'/> <br />
+The Single Shot MultiBox Detector (SSD)
+</p>


需要简要的解释下图中的SSD结构。

In this tutorial we used MobileNet. 这句话放在图的后面，引出下文。

qingqing01 · 2018-06-11T06:17:34Z

fluid/object_detection/README.md

-  ```
-  - Train on multi devices (/GPUs).
-
+train.py is the main caller of the training module. Examples of usage are shown below.


train.py -> `train.py`

qingqing01 · 2018-06-11T06:18:23Z

fluid/object_detection/README.md

  ```
+   - Set ```env CUDA_VISIBLE_DEVICES=0,1``` to specifiy the number of GPU you want to use.


env CUDA_VISIBLE_DEVICES=0,1
->
export CUDA_VISIBLE_DEVICES=0,1

qingqing01 · 2018-06-11T06:20:25Z

fluid/object_detection/README.md

  ```
+   - Set ```env CUDA_VISIBLE_DEVICES=0,1``` to specifiy the number of GPU you want to use.
+   - Set ```--dataset='coco2014'``` or ```--dataset='coco2017'``` to train model on MS COCO dataset.
+   - For more configurations, see the arguments list in train.py.


For more configurations, see the arguments list in train.py.

改成下面吧：

- For more help on arguments: ```bash python train.py --help ```

qingqing01 · 2018-06-11T06:22:15Z

fluid/object_detection/README.md

  ```python
-  env CUDA_VISIBLE_DEVICES=0,1 python -u train.py --batch_size=64 --dataset='pascalvoc' --pretrained_model='pretrained/ssd_mobilenet_v1_coco/'
+  python -u train.py --batch_size=64 --dataset='pascalvoc' --pretrained_model='pretrained/ssd_mobilenet_v1_coco/'


python -> bash

文档中所有命令行运行的地方，都改下吧。

qingqing01 · 2018-06-11T06:27:34Z

fluid/object_detection/README.md


 ### Infer and Visualize
-
+infer.py is the main caller of the infering module. Examples of usage are shown below.


`infer.py` is the main ...

qingqing01 · 2018-06-11T06:28:53Z

fluid/object_detection/README.md

+<img src="images/009943.jpg" height=300 width=400 hspace='10'/>
+<img src="images/009956.jpg" height=300 width=400 hspace='10'/>
+<img src="images/009960.jpg" height=300 width=400 hspace='10'/>
+<img src="images/009962.jpg" height=300 width=400 hspace='10'/> <br />


图像的类别name可以再调大点。

qingqing01 · 2018-06-11T06:30:52Z

fluid/object_detection/README.md


-Declaration: the MobileNet-v1 SSD model is converted by [TensorFlow model](https://github.com/tensorflow/models/blob/f87a58cd96d45de73c9a8330a06b2ab56749a7fa/research/object_detection/g3doc/detection_model_zoo.md). The MobileNet v1 model is converted [Caffe](https://github.com/shicai/MobileNet-Caffe).
+Declaration: the MobileNet-v1 SSD model is converted by [TensorFlow model](https://github.com/tensorflow/models/blob/f87a58cd96d45de73c9a8330a06b2ab56749a7fa/research/object_detection/g3doc/detection_model_zoo.md). The MobileNet v1 model is converted from [Caffe](https://github.com/shicai/MobileNet-Caffe).


这里的声明要么加上：我们将来会换成自己训练的模型？

qingqing01 · 2018-06-11T06:37:00Z

fluid/object_detection/README.md

 ```python
-env CUDA_VISIBLE_DEVICES=0 python infer.py --dataset='coco' --nms_threshold=0.5 --model_dir='train_coco_model/20' --image_path='./data/coco/val2014/COCO_val2014_000000000139.jpg'
+python infer.py --dataset='pascalvoc' --nms_threshold=0.45 --model_dir='train_pascal_model/best_model' --image_path='./data/pascalvoc/VOCdevkit/VOC2007/JPEGImages/009963.jpg'
 ```
 Below is the examples after running python infer.py to inference and visualize the model result.


Below is the examples after running python infer.py to inference and visualize the model result.

->

Below are the examples of running the inference and visualizing the model result.

qingqing01 · 2018-06-11T06:39:28Z

fluid/object_detection/README.md

@@ -125,5 +107,3 @@ TBD
 |MobileNet-v1-SSD 300x300  | COCO MobileNet SSD | VOC07+12 trainval| VOC07 test   | xx%  |


去掉表格里的最后两行吧，填下第一行的mAP结果。

qingqing01 · 2018-06-13T03:27:28Z

fluid/object_detection/README.md

+</p>
+
+SSD is readily pluggable into a wide variant standard convolutional network, such as VGG, ResNet, or MobileNet, which is also called base network or backbone. In this tutorial we used [MobileNet](https://arxiv.org/abs/1704.04861).
+

 ### Data Preparation

 You can use [PASCAL VOC dataset](http://host.robots.ox.ac.uk/pascal/VOC/) or [MS-COCO dataset](http://cocodataset.org/#download).

 #### PASCAL VOC Dataset


Remove #### PASCAL VOC Dataset.

qingqing01 · 2018-06-13T03:27:40Z

fluid/object_detection/README.md

@@ -25,7 +32,7 @@ The command `download.sh` also will create training and testing file lists.

 #### MS-COCO Dataset


Remove #### MS-COCO Dataset.

qingqing01 · 2018-06-13T03:30:46Z

fluid/object_detection/README.md

@@ -1,20 +1,27 @@
-The minimum PaddlePaddle version needed for the code sample in this directory is the lastest develop branch. If you are on a version of PaddlePaddle earlier than this, [please update your installation](http://www.paddlepaddle.org/docs/develop/documentation/en/build_and_install/pip_install_en.html).
+The minimum PaddlePaddle version needed for the code sample in this directory is the latest develop branch. If you are on a version of PaddlePaddle earlier than this, [please update your installation](http://www.paddlepaddle.org/docs/develop/documentation/en/build_and_install/pip_install_en.html).

 ---

 ## SSD Object Detection


Add Table of Contents before line 5, refer https://github.com/PaddlePaddle/models/pull/979/files#diff-e74a88ec056840ea1a6d958b791284afR5

qingqing01 · 2018-06-13T03:32:14Z

fluid/object_detection/README.md

  ```

-TBD
+We used RMSProp optimizer with mini-batch size 64 to train the MobileNet-SSD. The initial learning rate is 0.001, and was decayed at 40, 60, 80, 100 epochs with multiplier 0.5, 0.25, 0.1, 0.01, respectively. Weight decay is 0.00005. After 120 epochs we achive XXX% mAP under 11point metric.


Introduction the data argumentation and data reader before line 74. refer https://github.com/PaddlePaddle/models/pull/979/files#diff-e74a88ec056840ea1a6d958b791284afR87

update wiki

52a3c29

sefira force-pushed the ssd_wiki branch from b215b89 to 49cadd0 Compare June 7, 2018 07:44

refine wiki

a4fa39b

sefira force-pushed the ssd_wiki branch from 49cadd0 to a4fa39b Compare June 11, 2018 05:42

qingqing01 changed the title ~~SSD wiki~~ Update SSD documentation. Jun 11, 2018

qingqing01 reviewed Jun 11, 2018

View reviewed changes

sefira added 3 commits June 12, 2018 11:37

train.py default args

ec40950

update wiki cn

c207b3b

update wiki

2a20c0e

sefira force-pushed the ssd_wiki branch from ad16e55 to 2a20c0e Compare June 12, 2018 07:20

qingqing01 approved these changes Jun 13, 2018

View reviewed changes

qingqing01 merged commit 3aa16d5 into PaddlePaddle:develop Jun 13, 2018

qingqing01 reviewed Jun 13, 2018

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Update SSD documentation. #968

Update SSD documentation. #968

sefira commented Jun 7, 2018 •

edited by qingqing01

qingqing01 left a comment

qingqing01 Jun 11, 2018 •

edited

sefira Jun 11, 2018

qingqing01 Jun 11, 2018

sefira Jun 11, 2018

qingqing01 Jun 11, 2018

sefira Jun 11, 2018

qingqing01 Jun 11, 2018

sefira Jun 11, 2018

qingqing01 Jun 11, 2018

sefira Jun 11, 2018

qingqing01 Jun 11, 2018

sefira Jun 11, 2018 •

edited

qingqing01 Jun 11, 2018

sefira Jun 11, 2018

qingqing01 Jun 11, 2018

sefira Jun 11, 2018

qingqing01 Jun 11, 2018

sefira Jun 11, 2018

qingqing01 Jun 11, 2018

qingqing01 Jun 13, 2018

sefira Jun 15, 2018

qingqing01 Jun 13, 2018

sefira Jun 15, 2018

qingqing01 Jun 13, 2018

sefira Jun 15, 2018

qingqing01 Jun 13, 2018

sefira Jun 15, 2018

		```
		- Set ```env CUDA_VISIBLE_DEVICES=0,1``` to specifiy the number of GPU you want to use.


		### Infer and Visualize

		infer.py is the main caller of the infering module. Examples of usage are shown below.


		Declaration: the MobileNet-v1 SSD model is converted by [TensorFlow model](https://github.com/tensorflow/models/blob/f87a58cd96d45de73c9a8330a06b2ab56749a7fa/research/object_detection/g3doc/detection_model_zoo.md). The MobileNet v1 model is converted [Caffe](https://github.com/shicai/MobileNet-Caffe).
		Declaration: the MobileNet-v1 SSD model is converted by [TensorFlow model](https://github.com/tensorflow/models/blob/f87a58cd96d45de73c9a8330a06b2ab56749a7fa/research/object_detection/g3doc/detection_model_zoo.md). The MobileNet v1 model is converted from [Caffe](https://github.com/shicai/MobileNet-Caffe).

		@@ -125,5 +107,3 @@ TBD
		\|MobileNet-v1-SSD 300x300 \| COCO MobileNet SSD \| VOC07+12 trainval\| VOC07 test \| xx% \|

		@@ -25,7 +32,7 @@ The command `download.sh` also will create training and testing file lists.

		#### MS-COCO Dataset

Update SSD documentation. #968

Update SSD documentation. #968

Conversation

sefira commented Jun 7, 2018 • edited by qingqing01

qingqing01 left a comment

Choose a reason for hiding this comment

qingqing01 Jun 11, 2018 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sefira Jun 11, 2018 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sefira commented Jun 7, 2018 •

edited by qingqing01

qingqing01 Jun 11, 2018 •

edited

sefira Jun 11, 2018 •

edited