---
tags:
  - onnx
  - text-detection
  - textline_detection
  - computer-vision
  - japanese
  - ocr
  - OCR
library_name: tensorrt
language:
  - en
  - ja
pipeline_tag: image-to-text
---

mida text detector is the answer to the question: what would a text detection model look like that would be faster than meikiocr text detection, but more accurate than vita-vn-ocr textspot8 text detection?

inspiration:
- https://huggingface.co/rtr46/meiki.text.detect.v0
- https://github.com/bee-san/vita-vn-ocr

this model is experimental, trained a limited time on a limited training data set