Bottom up pose tracking demo

[HoBeom]

ADD Bottom up pose tracking demo
I need your review. 😄
Some docs may be misplaced. I'm sorry. 😢

1. ADD Bottom up pose tracking demo
2. I add an OKS similarity comparison.
3. One-Euro-Filter used by VNect is added for smoothing. 😮
   Performance is similar on the Posetrack 2018 VAL (HRNetw32_256x192).
   No filter : Total AP: 83.27367136220447
   filtered : Total AP, 83.29611050229873

• I separated this code from my complex code separately.
  This branch was made for push and was not tested.
  I believe your review will improve me.

[codecov]

Codecov Report

Merging #574 (43565a5) into master (a63b148) will decrease coverage by 0.28%.
The diff coverage is 66.33%.

@@            Coverage Diff             @@
##           master     #574      +/-   ##
==========================================
- Coverage   83.06%   82.78%   -0.29%     
==========================================
  Files         165      167       +2     
  Lines       11689    11963     +274     
  Branches     1881     1927      +46     
==========================================
+ Hits         9709     9903     +194     
- Misses       1518     1574      +56     
- Partials      462      486      +24     

Flag	Coverage Δ
unittests	82.78% <66.33%> (-0.20%)	⬇️

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

Impacted Files	Coverage Δ
mmpose/models/detectors/bottom_up.py	60.14% <ø> (ø)
mmpose/apis/inference_tracking.py	63.63% <39.28%> (-22.28%)	⬇️
mmpose/core/post_processing/__init__.py	100.00% <100.00%> (ø)
mmpose/core/post_processing/one_euro_filter.py	100.00% <100.00%> (ø)
mmpose/datasets/pipelines/shared_transform.py	81.50% <0.00%> (-7.00%)	⬇️
mmpose/datasets/pipelines/top_down_transform.py	62.42% <0.00%> (-1.19%)	⬇️
mmpose/apis/inference.py	48.41% <0.00%> (-1.19%)	⬇️
mmpose/datasets/pipelines/mesh_transform.py	94.31% <0.00%> (-1.14%)	⬇️
mmpose/models/detectors/top_down.py	58.15% <0.00%> (ø)
... and 3 more

---

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 a63b148...43565a5. Read the comment docs.

[jin-s13]

Thanks a lot for your contribution！We will take a look.

[ly015]

And could you please rebase your branch onto master? The demo folder has been reorganized in #592 so there are conflicts that need to be manually resolved.

[HoBeom]

I can't work right away because I'm on a business trip. I'll be working on it next week. 😭

[ly015]

Many thanks for the update, and we are almost done. Currently, the codecov rate is below the target, mainly due to the uncovered modification in apis/inference_tracking.py. Could you please add a piece of code into tests/test_api/test_inference_tracking.py to test the bottom-up tracking function (Copying the core part of bottom_up_pose_tracking_demo.py into test_inference_tracking should do it. You can follow the examples in test_inference_tracking.py) This would be very helpful for double-checking the function and future developments.

[HoBeom]

I wrote code to test the functions added in tracking by referring to tests/test_api/test_inference.py.