Contribution of Indian License Plate Dataset for v4.0 Accuracy

[VISHNU P]

Hello,

We are excited about the move to PyTorch and the architectural improvements in v4.0. To support better accuracy in the South Asian region, we are interested in providing a comprehensive dataset of Indian license plates.

To ensure the data is immediately useful for your new PyTorch-based pipeline, could you please clarify:

1. Annotation Format: Since you are moving to PyTorch, do you prefer YOLO (txt), COCO (json), or a specific XML format for bounding boxes and character labels?

2. Metadata Requirements: Should we include labels for vehicle make/model (VMMR) and color, or strictly focus on the plate and OCR characters?

3. Submission Method: What is your preferred method for transferring large datasets (e.g., Secure FTP, Google Drive link, or a specific repository)?

We look forward to contributing to the improvement of v4.0.

Regards,
Vishnu P
PETAERA TECHNOLOGIES LLP

[Mamadou DIOP]

Hi,

It's better sharing the whole images. We'll pass them through the detector to crop the plates just like how the inference pipeline would do. We'll automatically annotate the cropped plates using the current ocr code. Manual (human) correction is done on the errors. That's how we work.

For the training we use Pascal VOC format. 

So, we prefer doing the annotation by ourselves for the ocr part. If you have detection issues, we'll need Pascal voc annotations. 

You can put the images on Google drive and share the link with me or the support team (tech a-t doubango . org)

Sent from my Galaxy

--
You received this message because you are subscribed to the Google Groups "doubango-ai" group.
To unsubscribe from this group and stop receiving emails from it, send an email to doubango-ai...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/doubango-ai/f072d2c9-5ead-44cb-b595-8e5191c1e767n%40googlegroups.com.

[VISHNU P]

Hi,

As discussed earlier, we’ve shared the dataset privately via email from vis...@petaera.com to the support team, as suggested.

Just wanted to confirm whether it was received successfully, and please let us know if anything else is required from our side.

Regards,
Vishnu P
PETAERA TECHNOLOGIES LLP

[Mamadou DIOP]

Yes, they for it

To view this discussion visit https://groups.google.com/d/msgid/doubango-ai/ac51942d-ffbd-49c6-830e-1e36d836588bn%40googlegroups.com.

[VISHNU P]

Hi,

Thanks for confirming that the mail with the dataset was received.

Whenever convenient, I’d appreciate any feedback or response on the observations and issues related to ALPR version 3 that were mentioned in the same email. Please let me know if any further clarification or inputs are required from our side.

Looking forward to your response.

Regards,
Vishnu P
PETAERA TECHNOLOGIES LLP