GhiXu / ACMM

Multi-Scale Geometric Consistency Guided Multi-View Stereo
MIT License
155 stars 27 forks source link

ACMM input question #4

Closed Alihamdy2496 closed 2 years ago

Alihamdy2496 commented 2 years ago

Hi, thank you for providing this amazing work 🥇 🥇

please excuse ignorance , I am trying to use your method here with my SFM output , if I understood the script colmap2mvsnet_acm.py correctly the ACMM input is

+── input_folder
│   +── cams
│   │   +── pair.txt
│   │   +── 00000000_cam.txt
│   │   +── 00000001_cam.txt
│   │   +── 00000002_cam.txt
│   │   +── ...
│   +── images
│   │   +── 00000000.jpg
│   │   +── 00000001.jpg
│   │   +── 00000002.jpg
│   │   +── ...

the _cam.txt files are for example :

extrinsic
7.49715011e+01 -6.19671023e+01 -1.02564244e+03  7.84223389e+00
3.71642195e+02  8.19559173e+02 -2.30744366e+02  3.18972292e+00
8.65385352e-01 -7.98839231e-02 -4.94698647e-01  8.12122229e-03
0. 0. 0. 1.

intrinsic
853.36265647 0. 577.21334278
0. 852.01008808 370.2930352
0. 0. 1.

<depth_value1> <depth_value2> <depth_value3> <depth_value4> <<<< depth range and interval 

the pairs.txt file is for example :

 20 
 1
 <len(sorted_score>
 2
 <len(sorted_score>
 3
 <len(sorted_score>
 ....
 <sorted_score[image_id][0]>  <sorted_score[s][0]>
 <sorted_score[image_id][1]>  <sorted_score[s][1]>
 ...

my question is what are the 4 depth values range and interval ? and what is the sorted_score ?