Log In Sign Up
Models:
David Feaster
DavidFeaster/
Surgical-Bleeding-AMAGI
0
Downloads: 1
Diff of /tools/data/ucf101_24/README.md [000000] .. [6d389a]

Switch to unified view

a b/tools/data/ucf101_24/README.md 

  1
  
# Preparing UCF101-24





  
  

  2
  





  
  

  3
  
## Introduction





  
  

  4
  





  
  

  5
  
<!-- [DATASET] -->





  
  

  6
  





  
  

  7
  
```BibTeX





  
  

  8
  
@article{Soomro2012UCF101AD,





  
  

  9
  
  title={UCF101: A Dataset of 101 Human Actions Classes From Videos in The Wild},





  
  

  10
  
  author={K. Soomro and A. Zamir and M. Shah},





  
  

  11
  
  journal={ArXiv},





  
  

  12
  
  year={2012},





  
  

  13
  
  volume={abs/1212.0402}





  
  

  14
  
}





  
  

  15
  
```





  
  

  16
  





  
  

  17
  
For basic dataset information, you can refer to the dataset [website](http://www.thumos.info/download.html).





  
  

  18
  
Before we start, please make sure that the directory is located at `$MMACTION2/tools/data/ucf101_24/`.





  
  

  19
  





  
  

  20
  
## Download and Extract





  
  

  21
  





  
  

  22
  
You can download the RGB frames, optical flow and ground truth annotations from [google drive](https://drive.google.com/drive/folders/1BvGywlAGrACEqRyfYbz3wzlVV3cDFkct).





  
  

  23
  
The data are provided from [MOC](https://github.com/MCG-NJU/MOC-Detector/blob/master/readme/Dataset.md), which is adapted from [act-detector](https://github.com/vkalogeiton/caffe/tree/act-detector) and [corrected-UCF101-Annots](https://github.com/gurkirt/corrected-UCF101-Annots).





  
  

  24
  





  
  

  25
  
:::{note}





  
  

  26
  
The annotation of this UCF101-24 is from [here](https://github.com/gurkirt/corrected-UCF101-Annots), which is more correct.





  
  

  27
  
:::





  
  

  28
  





  
  

  29
  
After downloading the `UCF101_v2.tar.gz` file and put it in `$MMACTION2/tools/data/ucf101_24/`, you can run the following command to uncompress.





  
  

  30
  





  
  

  31
  
```shell





  
  

  32
  
tar -zxvf UCF101_v2.tar.gz





  
  

  33
  
```





  
  

  34
  





  
  

  35
  
## Check Directory Structure





  
  

  36
  





  
  

  37
  
After uncompressing, you will get the `rgb-images` directory, `brox-images` directory and `UCF101v2-GT.pkl` for UCF101-24.





  
  

  38
  





  
  

  39
  
In the context of the whole project (for UCF101-24 only), the folder structure will look like:





  
  

  40
  





  
  

  41
  
```





  
  

  42
  
mmaction2





  
  

  43
  
├── mmaction





  
  

  44
  
├── tools





  
  

  45
  
├── configs





  
  

  46
  
├── data





  
  

  47
  
│   ├── ucf101_24





  
  

  48
  
│   |   ├── brox-images





  
  

  49
  
│   |   |   ├── Basketball





  
  

  50
  
│   |   |   |   ├── v_Basketball_g01_c01





  
  

  51
  
│   |   |   |   |   ├── 00001.jpg





  
  

  52
  
│   |   |   |   |   ├── 00002.jpg





  
  

  53
  
│   |   |   |   |   ├── ...





  
  

  54
  
│   |   |   |   |   ├── 00140.jpg





  
  

  55
  
│   |   |   |   |   ├── 00141.jpg





  
  

  56
  
│   |   |   ├── ...





  
  

  57
  
│   |   |   ├── WalkingWithDog





  
  

  58
  
│   |   |   |   ├── v_WalkingWithDog_g01_c01





  
  

  59
  
│   |   |   |   ├── ...





  
  

  60
  
│   |   |   |   ├── v_WalkingWithDog_g25_c04





  
  

  61
  
│   |   ├── rgb-images





  
  

  62
  
│   |   |   ├── Basketball





  
  

  63
  
│   |   |   |   ├── v_Basketball_g01_c01





  
  

  64
  
│   |   |   |   |   ├── 00001.jpg





  
  

  65
  
│   |   |   |   |   ├── 00002.jpg





  
  

  66
  
│   |   |   |   |   ├── ...





  
  

  67
  
│   |   |   |   |   ├── 00140.jpg





  
  

  68
  
│   |   |   |   |   ├── 00141.jpg





  
  

  69
  
│   |   |   ├── ...





  
  

  70
  
│   |   |   ├── WalkingWithDog





  
  

  71
  
│   |   |   |   ├── v_WalkingWithDog_g01_c01





  
  

  72
  
│   |   |   |   ├── ...





  
  

  73
  
│   |   |   |   ├── v_WalkingWithDog_g25_c04





  
  

  74
  
│   |   ├── UCF101v2-GT.pkl





  
  

  75
  





  
  

  76
  
```





  
  

  77
  





  
  

  78
  
:::{note}





  
  

  79
  
The `UCF101v2-GT.pkl` exists as a cache, it contains 6 items as follows:





  
  

  80
  
:::





  
  

  81
  





  
  

  82
  
1. `labels` (list): List of the 24 labels.





  
  

  83
  
2. `gttubes` (dict): Dictionary that contains the ground truth tubes for each video.





  
  

  84
  
  A **gttube** is dictionary that associates with each index of label and a list of tubes.





  
  

  85
  
  A **tube** is a numpy array with `nframes` rows and 5 columns, each col is in format like `<frame index> <x1> <y1> <x2> <y2>`.





  
  

  86
  
3. `nframes` (dict): Dictionary that contains the number of frames for each video, like `'HorseRiding/v_HorseRiding_g05_c02': 151`.





  
  

  87
  
4. `train_videos` (list): A list with `nsplits=1` elements, each one containing the list of training videos.





  
  

  88
  
5. `test_videos` (list): A list with `nsplits=1` elements, each one containing the list of testing videos.





  
  

  89
  
6. `resolution` (dict): Dictionary that outputs a tuple (h,w) of the resolution for each video, like `'FloorGymnastics/v_FloorGymnastics_g09_c03': (240, 320)`.