Support upstream yolov5 P6 models

[zhiqwang]

We use the similar interface to load yolov5 P6 series models, which was first proposed in ultralytics/yolov5#2110 and released in ultralytics/yolov5#2762 , and we also support the latest version as described in ultralytics/yolov5#5141 .

from yolort.models import yolov5s6

# Load model
model = yolov5s6(pretrained=True, score_thresh=0.45)
model.eval()

# Perform inference on an image file
predictions = model.predict("bus.jpg")
# Perform inference on a list of image files
predictions = model.predict(["bus.jpg", "zidane.jpg"])

Note: We haven't added the pertained weights for P6 series models, and we will add these in a follow-up PR. (EDIT: Added in #206)

If you want load the custom checkpoint trained by ultralytics/yolov5, use the following snippets.

from yolort.models import YOLOv5

# 'yolov5s.pt' is downloaded from https://github.com/ultralytics/yolov5/releases/download/v6.0/yolov5s6.pt
ckpt_path_from_ultralytics = "yolov5s6.pt"
model = YOLOv5.load_from_yolov5(ckpt_path_from_ultralytics, version="r6.0")

model.eval()
img_path = "test/assets/bus.jpg"
predictions = model.predict(img_path)

This PR is bc-breaking, we deprecate the update_module_state_from_ultralytics and use load_from_ultralytics instead in ./yolort/utils/update_module_state.py.

https://github.com/zhiqwang/yolov5-rt-stack/blob/4b3454eb10f7cb4ef784036de01d0f79e72bee86/yolort/utils/update_module_state.py#L11-L20

Close #144 .

[codecov]

Codecov Report

Merging #196 (233a788) into master (ca3c97c) will increase coverage by 0.34%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #196      +/-   ##
==========================================
+ Coverage   96.49%   96.83%   +0.34%     
==========================================
  Files          10       10              
  Lines         599      601       +2     
==========================================
+ Hits          578      582       +4     
+ Misses         21       19       -2     

Flag	Coverage Δ
unittests	96.83% <100.00%> (+0.34%)	⬆️

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

Impacted Files	Coverage Δ
test/test_models.py	95.28% <100.00%> (-0.09%)	⬇️
test/test_utils.py	100.00% <100.00%> (+1.58%)	⬆️
test/test_v5.py	100.00% <0.00%> (+6.66%)	⬆️

---

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 ca3c97c...233a788. Read the comment docs.

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you all sign our Contributor License Agreement before we can accept your contribution.
1 out of 2 committers have signed the CLA.

✅ zhiqwang
❌ pre-commit-ci[bot]
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB