Dataset Directory

• color_image: RGB frames in PNG format with the frame rate of 60fps
• depth_raw: Depth frames in numpy format with the frame rate of 60fps, where the file name for each depth frame shares exactly same one with the file name of the corresponding RGB frame
• event_image: Event frames in PNG format with the frame rate of 120fps, which are generated by accumulating events in a time interval into an image. The synchronization of event frames with RGB and depth images can be found in the annotations
• event_raw: The raw original output files of the event camera. We have temporarily retained this file and NOT released yet
• color_label: Annotation files for RGB and depth images.
• event_label: Annotation files for event images

CDEHP
├── outdoor                             # Part of Dataset taken outdoors
│   ├── train                           # Train set
│   │   ├── A0001P0001S00               # A0001 -> Action 1, P0001 -> Person 1, S00 -> First Shot
│   │   │   ├── color_image             # RGB images
│   │   │   │   ├── 000000.png
│   │   │   │   ├── 000001.png
│   │   │   │   └── ******.png
│   │   │   ├── depth_raw               # Depth images
│   │   │   │   ├── 000000.npy
│   │   │   │   ├── 000001.npy
│   │   │   │   └── ******.npy
│   │   │   ├── event_image             # Event images
│   │   │   │   ├── 000000.png
│   │   │   │   ├── 000001.png
│   │   │   │   └── ******.png
│   │   │   ├── event_raw               # Raw event camera output bin file, NOT released yet
│   │   │   │   └── A0001P0001S00.bin
│   │   │   ├── color_label             # Label for RGB images
│   │   │   │   ├── 000000.txt
│   │   │   │   ├── 000001.txt
│   │   │   │   └── ******.txt
│   │   │   └── event_label             # Label for Event images
│   │   │       ├── 000000.txt
│   │   │       ├── 000001.txt
│   │   │       └── ******.txt
│   │   └── A0001P0002S00
│   └── valid                           # Validate set
│       └── ******
└── indoor                              # Part of Dataset taken indoors
    ├── train
    │   ├── A0001P0001S00
    │   └── ******
    └── valid

Definition of Keypoints

In our dataset, 13 keypoints were used to label the human body:

1. Head
2. Left Shoulder
3. Right Shoulder
4. Left Elbow
5. Right Elbow
6. Left Hand
7. Right Hand
8. Left Buttock
9. Right Buttock
10. Left Knee
11. Right Knee
12. Left Foot
13. Right Foot

Annotation File Format

Keypoints are annotated as a set of percentages with respect to the image width and length (x y). Please be noted that the resolution of RGB images and Event images are not same. RGB and Depth images are with the resolution of 848 x 480, Event images have a resolution of 1280 x 800.

Keypoint Labels on Event Frames

13                                      # Number of keypoints
event_name:000001                       # The file name of the corresponding event frame
0.7016509433962265 0.41458333333333336  # The keypoint coordinates, which are percentages with respect to the image width and length (x,y)
0.7158018867924528 0.4375
0.7099056603773585 0.4479166666666667
0.7264150943396226 0.5166666666666667
0.7169811320754716 0.5166666666666667
0.7252358490566038 0.5583333333333333
0.7075471698113207 0.5791666666666667
0.7122641509433962 0.5666666666666667
0.7287735849056604 0.5708333333333333
0.7087264150943396 0.6333333333333333
0.7488207547169812 0.6479166666666667
0.6886792452830188 0.7083333333333334
0.7594339622641509 0.71875

Keypoint Labels on Color and Depth Frames

13                                      # Number of keypoints
color_name:000000                       # the file names of the corresponding RGB and Depth frames
0.82513311428125 0.3706991523125        # The keypoint coordinates, which are percentages with respect to the image width and length (x,y)
0.8501153806093751 0.4064099953625
0.839520654015625 0.4227385711125
0.8702200228593749 0.53560962
0.8532573636249999 0.5353109638375
0.8686109479765625 0.6034388406375
0.8375879402109374 0.6384934238375
0.84581029965625 0.6178776785375
0.875207627203125 0.624133925625
0.840438244625 0.7273246002875
0.9124168345546876 0.7509422192875
0.8057340086406249 0.8516001396125
0.932000952484375 0.86591200105