[Feature]: Evaluation with TensorRT backend

[RunningLeon]

This PR includes:

• Add evaluation feature for TensorRT
• Evaluate supported models with TensorRT backend
• Update related docs.
• This PR depends on mmcv PR#1034 for FCOS model.

Results and Models

Model	Config	Metric	PyTorch	ONNX Runtime	TensorRT
FCOS	configs/fcos/fcos_r50_caffe_fpn_gn-head_4x4_1x_coco.py	Box AP	36.6	36.5	36.3
FSAF	configs/fsaf/fsaf_r50_fpn_1x_coco.py	Box AP	36.0	36.0	35.9
RetinaNet	configs/retinanet/retinanet_r50_fpn_1x_coco.py	Box AP	36.5	36.4	36.3
SSD	configs/ssd/ssd300_coco.py	Box AP	25.6	25.6	25.6
YOLOv3	configs/yolo/yolov3_d53_mstrain-608_273e_coco.py	Box AP	33.5	33.5	33.5
Faster R-CNN	configs/faster_rcnn/faster_rcnn_r50_fpn_1x_coco.py	Box AP	37.4	37.4	37.0
Mask R-CNN	configs/mask_rcnn/mask_rcnn_r50_fpn_1x_coco.py	Box AP	38.2	38.1	37.7
		Mask AP	34.7	33.7	33.3

Notes:

• All ONNX models are evaluated with dynamic shape on coco dataset and images are preprocessed according to the original config file.

• Mask AP of Mask R-CNN drops by 1% for ONNXRuntime. The main reason is that the predicted masks are directly interpolated to original image in PyTorch, while they are at first interpolated to the preprocessed input image of the model and then to original image in other backend.

• The performance drop for TensorRT is reasonable due to the detailed difference of resize op.

[codecov]

Codecov Report

Merging #5198 (abb83d8) into master (52c935d) will decrease coverage by 0.36%.
The diff coverage is 0.00%.

@@            Coverage Diff             @@
##           master    #5198      +/-   ##
==========================================
- Coverage   65.22%   64.86%   -0.37%     
==========================================
  Files         276      276              
  Lines       21240    21119     -121     
  Branches     3532     3534       +2     
==========================================
- Hits        13853    13698     -155     
- Misses       6635     6664      +29     
- Partials      752      757       +5     

Flag	Coverage Δ
unittests	64.85% <0.00%> (-0.34%)	⬇️

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

Impacted Files	Coverage Δ
mmdet/core/export/model_wrappers.py	0.00% <0.00%> (ø)
mmdet/models/detectors/single_stage.py	73.46% <0.00%> (+6.80%)	⬆️
mmdet/models/roi_heads/mask_heads/fcn_mask_head.py	69.41% <ø> (+0.66%)	⬆️
mmdet/models/dense_heads/rpn_test_mixin.py	77.41% <0.00%> (-6.46%)	⬇️
mmdet/models/utils/normed_predictor.py	88.09% <0.00%> (-4.77%)	⬇️
mmdet/core/anchor/point_generator.py	20.68% <0.00%> (-2.65%)	⬇️
mmdet/models/detectors/base.py	51.21% <0.00%> (-2.63%)	⬇️
mmdet/models/roi_heads/base_roi_head.py	85.29% <0.00%> (-2.21%)	⬇️
mmdet/datasets/pipelines/formating.py	64.22% <0.00%> (-2.16%)	⬇️
mmdet/core/bbox/coder/yolo_bbox_coder.py	58.97% <0.00%> (-2.01%)	⬇️
... and 51 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 52c935d...abb83d8. Read the comment docs.

[grimoire]

LGTM

[RunningLeon]

Ping @ZwwWayne