Downloads · 30 days
0
tungedng2710/tonai-license-plate-detection
tonai-license-plate-detection is a object detection model from tungedng2710. Use it when you need objects located in an image. It is set up for onnx. The card lists the license as mit.
licenseplatedetector.onnx is an ONNX object detection model for locating vehicle license plates in road-traffic imagery and video frames. The model is intended to be used as a detector in a larger automatic license pl…
Downloads · 30 days
0
Access
Public
Updated Jul 7, 2026
Repo size
12.1 MB
Likes
0
Public
Click a slice to open those files.
.onnx12.1 MB · 100%
From the Hugging Face model README
license_plate_detector.onnx is an ONNX object detection model for locating vehicle license plates in road-traffic imagery and video frames. The model is intended to be used as a detector in a larger automatic license plate recognition (ALPR) pipeline, where detected plate regions are cropped and passed to a downstream OCR or plate recognition model.
This model is intended for:
This model is not intended for:
license_plate_detector.onnx: ONNX-format license plate detector.The exact tensor names, input size, normalization, and output decoding should be verified from the ONNX graph before integration. Typical object detection usage requires:
NCHW.Use an ONNX runtime inspection tool to confirm the graph contract for your deployment environment.
import cv2
import numpy as np
import onnxruntime as ort
session = ort.InferenceSession("license_plate_detector.onnx", providers=["CPUExecutionProvider"])
input_name = session.get_inputs()[0].name
image = cv2.imread("vehicle.jpg")
# Adapt this preprocessing to the model's inspected input shape and training config.
input_shape = session.get_inputs()[0].shape
height, width = input_shape[2], input_shape[3]
resized = cv2.resize(image, (width, height))
tensor = resized[:, :, ::-1].astype(np.float32) / 255.0
tensor = np.transpose(tensor, (2, 0, 1))[None]
outputs = session.run(None, {input_name: tensor})
print(outputs)
The returned tensors must be decoded according to the exported detector architecture. Apply the same post-processing used during validation or export.
No standardized benchmark metrics are included with this upload. Before production use, evaluate the model on data that matches your target camera placement, geographic region, lighting, plate styles, weather, motion blur, and vehicle distribution.
Recommended evaluation metrics include:
License plate detection can be affected by:
This detector only localizes plate-like regions. It does not read plate text, verify plate validity, associate plates with identities, or provide legal conclusions.
Model performance may vary by geography, plate design, camera quality, road setting, and environmental conditions. Any deployment should include local validation, monitoring, and a process for human review where decisions may affect people.
License plates can be personal or regulated data depending on jurisdiction. Users are responsible for ensuring that collection, storage, processing, retention, and sharing comply with applicable privacy, data protection, surveillance, and transportation laws.
This model repository is released under the MIT License unless otherwise stated by the repository owner.