Update face detection (YuNet)

[cjpurackal]

Changes

Fixes #1781
Related to #2559

Type of change

• 📚 Documentation Update
• 🧪 Tests Cases
• 🐞 Bug fix (non-breaking change which fixes an issue)
• 🔬 New feature (non-breaking change which adds functionality)
• 🚨 Breaking change (fix or feature that would cause existing functionality to not work as expected)
• 📝 This change requires a documentation update

Checklist

• Support different configs
• ONNX compatible

[johnnv1]

can you add tests and docs updates? overall, it is fine, but we will break back compatibility here, add these changes, and deprecate the old version.

[edgarriba]

• would be interesting to see whether we can export to ONNX and torch compile the model
• Would be great too, to verify if there’s any speed performance difference

[johnnv1]

Any opinion about dataclass vs typed dict In these configs? related to #2093

[cjpurackal]

I think that makes sense 👍

[cjpurackal]

• would be interesting to see whether we can export to ONNX and torch compile the model

  • Would be great too, to verify if there’s any speed performance difference

I'm able to convert the model to ONNX. Inference works. I did some very basic tests with torch.compile and the resulting model was 10x slower. I'm unsure why. I'll investigate this later

[edgarriba]

@cjpurackal i think you need to rebase from main in order to fix the ci breaks