maianh511/PP_OCRv4_mobile_det_finetune_PortVehicleOCR

Model

PP-OCRv4_mobile_det was fine-tuned on the maianh511/PortVehicleOCR dataset — a collection of port vehicle images (license plates/containers) — to improve text detection performance on real-world images from this domain.

• Base model: PP-OCRv4_mobile_det — a lightweight text detection model optimized for edge deployment.
• Fine-tuned model: PP_OCRv4_mobile_det_finetune_PortVehicleOCR
• Dataset: maianh511/PortVehicleOCR — port vehicle images with text region annotations (license plates, container codes, etc.).

Objective

Improve text detection accuracy on domain-specific port/logistics images, which differ from the base model's general-purpose training data in lighting conditions, camera angles, and image quality.

Usage

from paddleocr import TextDetection

model = TextDetection(model_name="PP_OCRv4_mobile_det_finetune_PortVehicleOCR")
output = model.predict(input="your_image.jpg", batch_size=1)

for res in output:
    res.print()
    res.save_to_img(save_path="./output/")

Evaluation results (precision/recall/hmean) will be added once benchmarking on the test set is complete.