Make MMDetection config object accessible to users

[ai-fast-track]

I'm making the MMDetection config object accessible to users. The goal is to be able to update model attributes: e.g. changing weighted loss parameters, changing anchor boxes ratios, etc.

model_type = models.mmdet.retinanet
backbone = model_type.backbones.resnet50_fpn_1x
'cfg_options' = { 
  'model.bbox_head.loss_bbox.loss_weight': 2,
  'model.bbox_head.loss_cls.loss_weight': 0.8 }

# Passing cfg_options to the `model()` method to update loss weights
model = model_type.model(backbone=backbone(pretrained=True), num_classes=len(parser.class_map), cfg_options=cfg_options)

closes #903

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[codecov]

Codecov Report

Merging #904 (f2ad5d9) into master (2d91eac) will increase coverage by 0.01%.
The diff coverage is 100.00%.

❗ Current head f2ad5d9 differs from pull request most recent head fc3573a. Consider uploading reports for the commit fc3573a to get more accurate results

@@            Coverage Diff             @@
##           master     #904      +/-   ##
==========================================
+ Coverage   87.44%   87.45%   +0.01%     
==========================================
  Files         232      232              
  Lines        5011     5015       +4     
==========================================
+ Hits         4382     4386       +4     
  Misses        629      629              

Flag	Coverage Δ
unittests	87.45% <100.00%> (+0.01%)	⬆️

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

Impacted Files	Coverage Δ
...ion/models/mmdet/common/bbox/single_stage/model.py	100.00% <ø> (ø)
icevision/models/mmdet/common/utils.py	82.05% <100.00%> (+0.97%)	⬆️
icevision/models/mmdet/utils.py	68.62% <100.00%> (+1.28%)	⬆️

---

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 2d91eac...fc3573a. Read the comment docs.