[Feature] add codes for inference speed statistics

[heiyuxiaokai]

Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily get feedback. If you do not understand some items, don't worry, just make the pull request and seek help from maintainers.

Motivation

1. add codes for inference speed statistics from https://github.com/open-mmlab/mmdetection
2. fix the bug in eval_map

Modification

Please briefly describe what modification is made in this PR.

BC-breaking (Optional)

Does the modification introduce changes that break the back-compatibility of the downstream repos?
If so, please describe how it breaks the compatibility and how the downstream projects should modify their code to keep compatibility with this PR.

Use cases (Optional)

If this PR introduces a new feature, it is better to list some use cases here, and update the documentation.

Checklist

1. Pre-commit or other linting tools are used to fix the potential lint issues.
2. The modification is covered by complete unit tests. If not, please add more unit test to ensure the correctness.
3. The documentation has been modified accordingly, like docstring or example tutorials.

[yangxue0827]

This branch has conflicts that must be resolved

Conflicting files:
mmrotate/datasets/dota.py

You need to clone the dev branch and then submit the corresponding PR.
In addition, you need to add detailed docstring to all functions, you can refer to other scripts.

[heiyuxiaokai]

The inference speed of fp16 is about 1.5 times the speed of fp32 with a V100-SXM2-16GB. Here the gpu infos,

• FP16 (half) performance: 31.33 TFLOPS (2:1)
• FP32 (float) performance: 15.67 TFLOPS
• FP64 (double) performance: 7.834 TFLOPS (1:2)

It seems not to reach the limit.

[heiyuxiaokai]

This branch has conflicts that must be resolved

Conflicting files: mmrotate/datasets/dota.py

You need to clone the dev branch and then submit the corresponding PR. In addition, you need to add detailed docstring to all functions, you can refer to other scripts.

ok

[codecov-commenter]

Codecov Report

Merging #86 (dfcea3f) into dev (4e39a45) will increase coverage by 0.28%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##              dev      #86      +/-   ##
==========================================
+ Coverage   28.67%   28.96%   +0.28%     
==========================================
  Files         102      104       +2     
  Lines        6604     6581      -23     
  Branches      990      985       -5     
==========================================
+ Hits         1894     1906      +12     
+ Misses       4646     4609      -37     
- Partials       64       66       +2     

Flag	Coverage Δ
unittests	28.94% <ø> (+0.28%)	⬆️

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

Impacted Files	Coverage Δ
...oi_extractors/rotate_single_level_roi_extractor.py	23.80% <0.00%> (-0.79%)	⬇️
mmrotate/models/dense_heads/rotated_anchor_head.py	10.12% <0.00%> (-0.64%)	⬇️
mmrotate/core/__init__.py	100.00% <0.00%> (ø)
mmrotate/apis/inference.py	0.00% <0.00%> (ø)
mmrotate/datasets/pipelines/loading.py	30.43% <0.00%> (ø)
...ate/models/dense_heads/kfiou_rotate_retina_head.py	30.00% <0.00%> (ø)
...e/models/dense_heads/rotated_retina_refine_head.py	25.00% <0.00%> (ø)
...els/dense_heads/kfiou_rotate_retina_refine_head.py	25.00% <0.00%> (ø)
mmrotate/core/evaluation/eval_map.py	7.58% <0.00%> (ø)
mmrotate/core/evaluation/__init__.py	100.00% <0.00%> (ø)
... and 10 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 7c9a592...dfcea3f. Read the comment docs.

[yangxue0827]

Thank you for your contributions. Please follow https://github.com/open-mmlab/mmcv/blob/master/CONTRIBUTING.md to set pre-commit and fix the lint error.

[yangxue0827]

All checks have passed, looking forward to the addition of the docstring.

[yangxue0827]

LGTM

[jbwang1997]

LGTM

[zytx121]

LGTM