0

I basically wanna tell Google Vision to read from top to bottom, left to right. In this example you can see the part on top being recognized as last instead of first. I guess, i could painfully manually go through all the vertices and check coords, but I'd like to avoid that if possible. I've been searching in the docs, but couldn't find anything and/or simply don't know enough about OCR in general.

google vision ocr example

const r = await vs.batchAnnotateImages({ requests: [
        {
        features: [{type: 'TEXT_DETECTION'}],
        image: { content: base64String },
        imageContext: {
            languageHints: ["ja"]
        },
    }]})
Otm Shank
  • 1
  • I'm currently having this problem too. Sadly, from the [document](https://cloud.google.com/vision/docs/reference/rest/v1/AnnotateImageRequest), there is no settings in `features`, `imageContext` properties. I have to painfully manually go through all the vertices and check coords. – vee Dec 04 '22 at 17:12
  • From [this answer](https://stackoverflow.com/a/48279030/128761), I've copy to **web browser version** [here, on gitHub](https://github.com/Rundiz/line-segmentation-algorithm-to-gcp-vision-webbrowser). Hope this help for anyone who is looking for Google Vision text direction problem. – vee Dec 06 '22 at 03:39

0 Answers0