manually_download_datasets.md

Manually download iPER and MotionSynthetic datasets

All available download links:

• OneDrive 1

• OneDrive 2

• BaiduPan, password: uomm

Download iPER dataset

In the local disk, the iPER root folder is $iPER_root_dir, and cd to $iPER_root_dir.

1. download train.txt, val.txt, and move them into $iPER_root_dir

2. download all split files from splits/images_HD.tar.gz.aa to splits/images_HD.tar.gz.bk, and move them to $iPER_root_dir/splits, and we will arrive at the following file structure:

   $iPER_root_dir:
       |-- splits
       |   |-- images_HD.tar.gz.aa
       |   |-- images_HD.tar.gz.ab
       |   |-- images_HD.tar.gz.ac
       |   |-- ......
       |   |-- images_HD.tar.gz.bj
       |   `-- images_HD.tar.gz.bk
       |-- train.txt
       `-- val.txt

3. merge all the splits files into images_HD.tar.gz and move it to $iPER_root_dir,

   cat splits/images_HD.tar.gz.*   >   images_HD.tar.gz

   Then, we will arrive at the following file structure

   $iPER_root_dir:
       |-- splits
       |   |-- images_HD.tar.gz.aa
       |   |-- images_HD.tar.gz.ab
       |   |-- images_HD.tar.gz.ac
       |   |-- ......
       |   |-- images_HD.tar.gz.bj
       |   `-- images_HD.tar.gz.bk
       |-- images_HD.tar.gz
       |-- train.txt
       `-- val.txt

4. extract images_HD.tar.gz to $iPER_root_dir,

   tar  -xzvf images.tar.gz

   Then, we will arrive at the following file structure

   $iPER_root_dir:
       |-- images_HD
       |   |-- 001
       |   |   |-- 1
       |   |   |   |-- 1
       |   |   |   `-- 2
       |   |   |-- 10
       |   |   |   |-- 1
       |   |   |   `-- 2
       |   |   |-- 11
       |   |   |   |-- 1
       |   |   |   `-- 2
       |   |   |-- 12
       |  .......
       |   |-- 029
       |   |   `-- 1
       |   |   |   |-- 1
       |   |   |   `-- 2
       |   `-- 030
       |       `-- 1
       |   |   |   |-- 1
       |   |   |   `-- 2
       |-- splits
       |   |-- images_HD.tar.gz.aa
       |   |-- images_HD.tar.gz.ab
       |   |-- images_HD.tar.gz.ac
       |   |-- ......
       |   |-- images_HD.tar.gz.bj
       |   `-- images_HD.tar.gz.bk
       |-- images_HD.tar.gz
       |-- train.txt
       `-- val.txt

5. run the following script to process the iPER dataset

   python scripts/train/prepare_iPER_dataset.py \
      --output_dir $iPER_root_dir \
      --gpu_ids 0,0,0,1,1,1,2,2,2,3,3,3,5,5,5,6,6,6,7,7,7,8,8,8,9,9,9

   Then, all the processed information will be saved into $iPER_root_dir/primitives, and we will arrive at the following file structure,

    $iPER_root_dir:
     |-- images_HD
     |   |-- 001
     |   |   |-- 1
     |   |   |   |-- 1
     |   |   |   `-- 2
     |   |   |-- 10
     |   |   |   |-- 1
     |   |   |   `-- 2
     |   |   |-- 11
     |   |   |   |-- 1
     |   |   |   `-- 2
     |   |   |-- 12
     |  .......
     |   |-- 029
     |   |   `-- 1
     |   |   |   |-- 1
     |   |   |   `-- 2
     |   `-- 030
     |       `-- 1
     |   |   |   |-- 1
     |   |   |   `-- 2
     |-- splits
     |   |-- images_HD.tar.gz.aa
     |   |-- images_HD.tar.gz.ab
     |   |-- images_HD.tar.gz.ac
     |   |-- ......
     |   |-- images_HD.tar.gz.bj
     |   `-- images_HD.tar.gz.bk
     |-- images_HD.tar.gz
     |-- primitives
     |   |-- 001
     |   |   |-- 1
     |   |   |   |-- 1
     |   |   |   `-- 2
     |   |   |-- 10
     |   |   |   |-- 1
     |   |   |   `-- 2
     |   |-- 005
     |   |   `-- 1
     |   |       |-- 1
     |   |       `-- 2
     |  ......
     |   `-- 030
     |       `-- 1
     |           |-- 1
     |           `-- 2
     |-- train.txt
     `-- val.txt

Download motionSynthetic dataset

The MotionSynthetic root folder is $MotionSynthetic_root_dir, and cd to $MotionSynthetic_root_dir.

1. download train.txt and val.txt, and move them to $MotionSynthetic_root_dir;

2. download videos.zip and poses.zip, and move them to $MotionSynthetic_root_dir;

3. unzip videos.zip and poses.zip

   unzip videos.zip
   unzip poses.zip

   Then, we will arrive at the following file structure:

   $MotionSynthetic_root_dir
   |-- poses
   |   |-- MG_125611487366942_0366_0018
   |   |   |-- kps.pkl
   |   |   `-- pose_shape.pkl
   |   |-- MG_125611494277906_2031_0017
   |   |   |-- kps.pkl
   |   |   `-- pose_shape.pkl
   |   | ......
   |   `-- PeopleSnapshot_male-9-plaza_2211_0016
   |       |-- kps.pkl
   |       `-- pose_shape.pkl
   |-- poses.zip
   |-- train.txt
   |-- val.txt
   |-- videos
   |   |-- MG_125611487366942_0366_0018.mp4
   | ......
   |   `-- PeopleSnapshot_male-9-plaza_2211_0016.mp4
   `-- videos.zip
   

4. run the following script to process the MotionSynthetic dataset

   python scripts/train/prepare_motionSynthetic_dataset.py \
      --output_dir /p300/tpami/datasets_reproduce/motionSynthetic \
      --gpu_ids 0,0,0,0,1,1,1,1

   Then, all the processed information will be saved into $MotionSynthetic_root_dir/primitives, and we will arrive at the following file structure:

   $MotionSynthetic_root_dir
   |-- poses
   |   |-- MG_125611487366942_0366_0018
   |   |   |-- kps.pkl
   |   |   `-- pose_shape.pkl
   |   |-- MG_125611494277906_2031_0017
   |   |   |-- kps.pkl
   |   |   `-- pose_shape.pkl
   |   | ......
   |   `-- PeopleSnapshot_male-9-plaza_2211_0016
   |       |-- kps.pkl
   |       `-- pose_shape.pkl
   |-- poses.zip
   |-- primitives
   |   |-- MG_125611487366942_0366_0018
   |   |   `-- processed
   |   |       |-- actual_background
   |   |       |-- background
   |   |       |-- images
   |   |       |-- parse
   |   |       |-- vid_info.pkl
   |   |       `-- visual.mp4
   |   | ......
   |   `-- PeopleSnapshot_male-9-plaza_2211_0016
   |       `-- processed
   |           |-- actual_background
   |           |-- background
   |           |-- images
   |           |-- parse
   |           |-- vid_info.pkl
   |           `-- visual.mp4
   |-- train.txt
   |-- val.txt
   |-- videos
   |   |-- MG_125611487366942_0366_0018.mp4
   | ......
   |   `-- PeopleSnapshot_male-9-plaza_2211_0016.mp4
   `-- videos.zip