Skip to content

Latest commit

 

History

History
97 lines (65 loc) · 3.15 KB

data_type.md

File metadata and controls

97 lines (65 loc) · 3.15 KB
Raw

RGB Image

The color images are stored as 640x480 8-bit RGB images in PNG format.

  • Load the image using OpenCV:
import cv2
img = cv2.imread(FILENAME)
cv2.imshow('img', img)
cv2.waitKey(0)
  • Load the image using Pillow:
from PIL import Image
img = Image.open(FILENAME)
img.show()

Camera intrinsics

fx = 772.5483399593904  # focal length x
fy = 772.5483399593904  # focal length y
cx = 320.0  # optical center x
cy = 180.0  # optical center y

fov = 45 deg # field of view

width = 640
height = 360

Depth image

The depth maps are stored as 640x480 16-bit numpy array in NPY format. In the Unreal Engine, the environment usually has a sky sphere at a large distance. So the infinite distant object such as the sky has a large depth value (e.g. 10000) instead of an infinite number.

The unit of the depth value is meter. The baseline between the left and right cameras is 0.25m.

  • Load the depth image:
import numpy as np
depth = np.load(FILENAME)

# change to disparity image
disparity = 80.0 / depth

RCNN Segmentation image

The segmentation masks are saved as a uint8 array. the shape of the images is equal to the images' shape. The index in the corresponding pixel represent the object category (0 menas background).

  • Load the segmentation image:
import cv2
depth = cv2.imread(FILENAME, cv2.IMREAD_GRAYSCALE)

Alphapose file

The human pose file is a text file containing the 2D position on image plane (in pixel). Each file is corresponding to one image frame in the sequence.

  • Each line in the file represents a single detected human.

  • The format of the file is 'N * (u_x, u_y, p)' (18 * 3)

  • u_x, uy represent the x and y coordinate of the human key points.

  • p represent the probability of the detected human key points.

Track id file

The tracking id of the detected human. Note this is a referenced tracking result genrerated by nerarest neighbor searching.

  • The id is unique, each line is corresponding to the detected human in alphapose_0

  • -1 is assigned if the tracking is lost. A new object id will be assigned in this case.

Pose file

The camera pose file is a text file containing the translation and orientation of the camera in a fixed coordinate frame. Note that our automatic evaluation tool expects both the ground truth trajectory and the estimated trajectory to be in this format.

  • Each line in the text file contains a single pose.

  • The number of lines/poses is the same as the number of image frames in that trajectory.

  • The format of each line is 'time_stamp tx ty tz qx qy qz qw'.

  • tx ty tz (3 floats) give the position of the optical center of the color camera with respect to the world origin in the world frame.

  • qx qy qz qw (4 floats) give the orientation of the optical center of the color camera in the form of a unit quaternion with respect to the world frame.

  • The camera motion is defined in the NED frame. That is to say, the x-axis is pointing to the camera's forward, the y-axis is pointing to the camera's right, the z-axis is pointing to the camera's downward.

  • Load the pose file:

import numpy as np
flow = np.loadtxt(FILENAME)