[Feature] Support online visualization of semantic segmentation results

[Wuziyi616]

Similar to detection online visualization, I use open3d to show segmentation masks. The main difference here is that, in det, we can draw gt and pred bbox on top of scene_points, because they won't overlap too much with each other. However the drawing of seg requires per-point colorization, while scene_points often have their own colors. So we can't draw them together.

My solution is that, I set an offset for each seg mask and then draw them in a row. See examples below.

[Wuziyi616]

The saved offline results (on ScanNet dataset, the pred_mask is randomly generated):

Points:

GT:

Pred:

Online visualization:

[codecov]

Codecov Report

Merging #416 (9970aa4) into master (9cb75e7) will decrease coverage by 0.08%.
The diff coverage is 18.51%.

@@            Coverage Diff             @@
##           master     #416      +/-   ##
==========================================
- Coverage   50.70%   50.62%   -0.09%     
==========================================
  Files         182      182              
  Lines       12765    12788      +23     
  Branches     2014     2019       +5     
==========================================
+ Hits         6473     6474       +1     
- Misses       5875     5895      +20     
- Partials      417      419       +2     

Flag	Coverage Δ
unittests	50.62% <18.51%> (-0.09%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
mmdet3d/core/visualizer/open3d_vis.py	0.00% <0.00%> (ø)
mmdet3d/core/visualizer/show_result.py	67.03% <33.33%> (-7.97%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 9cb75e7...9970aa4. Read the comment docs.