v1_layout.md

V1

We provide folders with different data types. As new versions of the dataset are released, they may have different formats available.

Data type	Description
rgb	RGB images
mask	Binary mask of foot
normals	Predicted normals
labels	JSON files with additional information (more info)

JSON format

We provide preliminary data that might be useful in our labels.json

{
  "camera" : {
    "pos": [0, 0, 0], # world camera position
    "rot": [0, 0, 0], # world camera rotation (euler, radians)
    "fov": 60.0 # camera field of view (degrees)
  },
  
  "foot" : {
      "name": "0009-A" # Foot3D scan used   
  },
  
  "keypoints" : {
      "big toe": [0, 0], # pixel coordinates
      ...
  }
}

Keypoints available are: ['big toe', '2nd toe', '3rd toe', '4th toe', 'little toe', 'heel', 'outer extrema', 'inner extrema']