sample feicuiwan images:180 points:132070 image_size:6144*1096
yingrenshi_downsample image:855 points:~660000 image_size:2736 × 1824
to run the large scene calibration imutils should be installed timm also
the large scene preprocessing contain 5 step(dataset do not contain the gps)
- feature extraction
- use vocal tree to match the feature
- mapper
- calibration the coordinate
- use midas model to calculate the depth map