Request for Assistance: Validation Dataset for Tesseract Model (Mercosul License Plates)

[Guilherme Alves]

Hello everyone,

I’m currently working on fine-tuning a Tesseract OCR model to improve its accuracy in recognizing  Mercosul license plates  (e.g., Brazilian, Argentinean formats). While I’ve made progress with training data, I’m struggling to find a  validation dataset  containing real-world images of these plates to properly evaluate the model’s performance.

Would anyone in this community be willing to share a dataset or point me to resources where I might obtain one? I’d greatly appreciate any help, even a small sample of images.

[Mike]

Hi, 

Have you tried applying augmentation to your training data? You might be able to siphon enough off for a training set. 

[abhimanyu yadav]

plz contact me on abhi...@parkmont.co.in