Merge branch 'master' of https://github.com/AetrexTechnology/mmsegmentation

Author: ZhimingNJ

docs/en/dataset_prepare.md

@@ -19,8 +19,7 @@ mmsegmentation
 │   │   ├── val_set
 │   │   │   ├── images
 │   │   │   ├── annotations
-│   |   ├── train_list.txt
-│   │   ├── val_list.txt
+│   |   |
 
 ```
 

docs/en/inference.md

@@ -1,6 +1,6 @@
 ## Inference with pretrained models
 
-The testing script 'arch_segmentation.py' is in folder demo. We can run the following commands to obtain the arch boundary in a color image. 
+The testing script 'arch_segmentation.py' is in the folder demo. We can run the following commands to obtain the arch boundary in a color image. 
 
 ```shell
 python demo/arch_segmentation.py ${CONFIG_FILE} ${CHECKPOINT_FILE} --img-path ${INPUT_DIR} --show-dir ${OUTPUT_DIR} 
@@ -23,9 +23,18 @@ Assume that you run the script in Windows system.
 ```shell
 python demo/arch_segmentation.py C:/Users/zliu/Projects/configs/mmseg_configs/models/point_rend_r101_512x512_160k_feetarch5k.py \
                                  C:/Users/zliu/Projects/models/mmseg/arch_pointrend/iter_2000.pth \
-                                 --img-path C:/Users/zliu/Projects/foot_imgs/val_set/images
-                                 --show-dir C:/Users/zliu/Projects/foot_imgs/show
-                                 --palette feetarch
-                                 --opacity 0.4
+                                 --img-path C:/Users/zliu/Projects/foot_imgs/val_set/images \
+                                 --show-dir C:/Users/zliu/Projects/foot_imgs/show \
+                                 --palette feetarch \
+                                 --opacity 0.4 
 ```
 
+Some segmentation results are shown below:
+
+![arch1](../../resources/feet_imgs/Ted_scan4_right_front_arch.png)
+![arch2](../../resources/feet_imgs/Tony_R_scan2_right_back_arch.png)
+![arch3](../../resources/feet_imgs/Wendy_scan4_right_front_arch.png)
+![arch3](../../resources/feet_imgs/Whitney_scan3_left_back_arch.png)
+![arch3](../../resources/feet_imgs/Yanina_scan2_left_back_arch.png)
+
+A pickle file 'output.pkl' is generated in the folder ${OUTPUT_DIR}, which contains the list of numpy 2d array, each encoding the arch boundary.