craft-mlt-25k-onnx

craft-mlt-25k-onnx

ONNX version of CRAFT (Character Region Awareness for Text Detection), an image-based text detection model.

Installation

pip install torch transformers optimum-onnx onnxruntime

For GPU inference, install a CUDA-compatible PyTorch build and replace onnxruntime with the compatible onnxruntime-gpu package. Do not install both ONNX Runtime packages in the same environment.

Usage

import onnxruntime as ort
from huggingface_hub import hf_hub_download

model_path = hf_hub_download(
    repo_id="KvaytG/craft-mlt-25k-onnx",
    filename="craft.onnx",
)

available_providers = ort.get_available_providers()
providers = (
    ["CUDAExecutionProvider", "CPUExecutionProvider"]
    if "CUDAExecutionProvider" in available_providers
    else ["CPUExecutionProvider"]
)

session = ort.InferenceSession(model_path, providers=providers)

print("Input:", session.get_inputs()[0].name, session.get_inputs()[0].shape)
print("Outputs:", [(output.name, output.shape) for output in session.get_outputs()])

Model Details

• Base model: CRAFT (Character Region Awareness for Text Detection).
• Format: ONNX.
• ONNX opset: 18.
• Inference engine: ONNX Runtime.
• Export framework: Hugging Face Optimum.

Credits

• Original implementation: CRAFT (Character Region Awareness for Text Detection).
• ONNX export framework: Hugging Face Optimum
• Inference runtime: ONNX Runtime

License

Licensed under the MIT license, following the license of the original model.

Citation

@misc{kvaytg_craft_mlt_25k_onnx,
  author       = {KvaytG},
  title        = {CRAFT MLT 25K ONNX},
  year         = {2026},
  publisher    = {Hugging Face},
  journal      = {Hugging Face Models},
  url          = {https://huggingface.co/KvaytG/craft-mlt-25k-onnx},
  note         = {ONNX export of CRAFT}
}