Hello, thanks for the great work !
I am trying to generate a sample of this dataset with the matterport3d data.
The outputs I am getting seem to be low quality. Is this a bug in my visualization or is this expected? Do you rely that with sufficiently large amounts of data the noise will cancel out ?
Examples:
Objects (11): towel_0, sink, desk, mirror_0, mirror_1, towel_1, bottle, tissue, cabinet, counter, light
Anchor [1/6]: sink
Frame: cameracentric
infront (red): 19 pts
behind (blue): 16 pts
left (lime): 49 pts
right (yellow): 0 pts
Image: matterport3d/1pXnuDYAj8r/region18/5a72c23b99fa40f1bcd9dfbb60340b05_i1_2.jpg
Objects (7): window, chair_0, desk, curtain, doorframe, chair_1, pad
Anchor [1/2]: window
Frame: cameracentric
infront (red): 5 pts
behind (blue): 0 pts
left (lime): 5 pts
right (yellow): 0 pts
Image: matterport3d/1pXnuDYAj8r/region13/1407cf2da95e4f73951798295db82ab6_i1_0.jpg
Objects (32): oven_0, kitchen island, table, counter_0, decoration, teapot, chair_0, cabinet_0, counter_1, chair_1, cabinet_1, chair_2, stove, cabinet_2, bottle, chair_3, plate_0, range hood, plate_1, oven_1, chair_4, counter_2, plate_2, cabinet_3, chair_5, faucet, bowl, chair_6, plate_3, chair_7, cabinet_4, doorframe
Anchor [1/3]: kitchen island
Frame: cameracentric
infront (red): 21 pts
behind (blue): 0 pts
left (lime): 0 pts
right (yellow): 50 pts
Also cameracentric and worldcentric seem to have identical points for unary spatial questions. Is this expected ?
Thanks again !
Hello, thanks for the great work !
I am trying to generate a sample of this dataset with the matterport3d data.
The outputs I am getting seem to be low quality. Is this a bug in my visualization or is this expected? Do you rely that with sufficiently large amounts of data the noise will cancel out ?
Examples:
Also cameracentric and worldcentric seem to have identical points for unary spatial questions. Is this expected ?
Thanks again !