Raise error on suffix-less model path

[democat3457]

This PR raises an Exception when the model passed into DetectMultiBackend does not have a suffix.

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Enforced format support validation in YOLOv5 TensorFlow model conversion.

📊 Key Changes

• Added an error raise condition for unsupported format checks during model conversion in TensorFlow.

🎯 Purpose & Impact

• Purpose: Ensures that the user is informed if they attempt to use an unsupported export format when converting YOLOv5 models using the TensorFlow wrapper.
• Impact: Provides clearer error messaging, preventing potential confusion and troubleshooting time for users. It ensures users only work with compatible formats, enhancing overall user experience with the model conversion process. 🛠️🔍

[glenn-jocher]

@democat3457 the weight suffix is checked here on L520 against the valid ones and an error is thrown if it doesn't match:

yolov5/models/common.py

Lines 516 to 526 in 574ceed

	def model_type(p='path/to/model.pt'):
	# Return model type from model path, i.e. path='path/to/model.onnx' -> type=onnx
	from export import export_formats
	suffixes = list(export_formats().Suffix) + ['.xml'] # export suffixes
	check_suffix(p, suffixes) # checks
	p = Path(p).name # eliminate trailing separators
	pt, jit, onnx, xml, engine, coreml, saved_model, pb, tflite, edgetpu, tfjs, xml2 = (s in p for s in suffixes)
	xml |= xml2 # *_openvino_model or *.xml
	tflite &= not edgetpu # *.tflite
	return pt, jit, onnx, xml, engine, coreml, saved_model, pb, tflite, edgetpu, tfjs

So this PR initially looked like a good idea but I think it's redundant with the above check.

[democat3457]

I see, it seems like the case I had was that the string I passed in had no suffix, so the assert wasn't even being checked - I'll update the check suffix method instead

[glenn-jocher]

@democat3457 ah yes I just remembered! The check_suffix() method has no suffix as one of the acceptable options in case the user passes a directory, like for OpenVINO models. So I think your original PR is probably good. Can you revert the last commit and I'll go ahead and merge? Thanks!

[democat3457]

oh oops, ok then

[glenn-jocher]

@democat3457 PR is merged. Thank you for your contributions to YOLOv5 🚀 and Vision AI ⭐