做了一个 ocr 的网站，把图片转为文本，可以看到文本对应的位置

基于什么技术呢？
现在开源的很多，拉一个容器就有了。

太慢了哈哈

各种开源成品拿来就用，就算定制的也就自己找样本识别、训练一下就完了，你这有什么优势？

有问题啊，怎么发图片 我就上传一张 经营许可证 全乱了 这里怎么发图

@xiao8276 点 choose file 就可以哈

@249239432 在一些数据上做过微调，效果会更好一点，除此之外没优势

确实慢

@Seanfuck 基于 paddleocr 模型的微调

@cooljser 图片大的话会比较慢，在 cpu 上预估的，后面会做一些加速

样式全乱了，加油

data:image/jpeg;base64,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

@japhetJiu 感谢反馈，图片解码图来的图裂了

@xiao8276 移动端适配有问题

你就直接说拿他模型搭建的不就完了，
样本数据要么自己训练，要么用别人的，什么叫一些数据上微调？

@249239432 你说的都对

都 500 错误了也不提示，还在那儿 Loading...

orc 不是兽族么，是 ocr 吧