Refactor train.py/test.py

[wHao-Wu]

[Enhance] Refactoring the train.py/test.py

• Changing details in train.py

  • Re-seting gpu_ids with distributed training mode
  • Adding "cfg-options" and deprecating "options" in arguments
  • Adding support for "custom_imports"

• Changing details in test.py

  • Changing the import path of test.py
    • Switching wrap_fp16_model from path "mmcv" to path "mmdet"
  • Adding support for "custom_imports"
  • Adding "cfg-options" and "eval-options", and deprecating "options" in argument
  • Adding support when test dataset is concatenated
  • Changing op when args.fuse_conv_bn is true
  • Changing interface in evaluation

• Verifying refactor above with SECOND model

[codecov]

Codecov Report

Merging #343 (6370199) into master (e366272) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master     #343   +/-   ##
=======================================
  Coverage   49.69%   49.69%           
=======================================
  Files         174      174           
  Lines       11758    11758           
  Branches     1838     1838           
=======================================
  Hits         5843     5843           
  Misses       5555     5555           
  Partials      360      360           

Flag	Coverage Δ
unittests	49.69% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out 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 e366272...f3aa185. Read the comment docs.