hub server部署不支持 batch_size 参数

[BobDu]

根据readme config.json

{
  "modules_info": {
    "chinese_ocr_db_crnn_mobile": {
      "init_args": {
        "version": "1.0.3"
      },
      "predict_args": {
        "use_gpu": true,
        "batch_size": 20
      }
    }
  },
  "port": 3306
}

服务报错

2020-08-21 20:08:55,054-INFO: 211.103.135.176 - - [21/Aug/2020 20:08:55] "POST /predict/chinese_ocr_db_crnn_mobile HTTP/1.1" 200 -
2020-08-21 20:09:08  -  recognize_text() got an unexpected keyword argument 'batch_size'

阅读paddlehub的源码中

~/Projects/PaddleHub/hub_module/modules/image/text_recognition/chinese_ocr_db_crnn_mobile/module.py

    def recognize_text(self,
                       images=[],
                       paths=[],
                       use_gpu=False,
                       output_dir='ocr_result',
                       visualization=False,
                       box_thresh=0.5,
                       text_thresh=0.5):
        """
        Get the chinese texts in the predicted images.
        Args:
            images (list(numpy.ndarray)): images data, shape of each is [H, W, C]. If images not paths
            paths (list[str]): The paths of images. If paths not images
            use_gpu (bool): Whether to use gpu.
            batch_size(int): the program deals once with one
            output_dir (str): The directory to store output images.
            visualization (bool): Whether to save image or not.
            box_thresh(float): the threshold of the detected text box's confidence
            text_thresh(float): the threshold of the recognize chinese texts' confidence
        Returns:
            res (list): The result of chinese texts and save path of images.
        """

recognize_text 方法并没有按照文档中写的接受batch_size参数

如果生产化部署中没有 batch_size 预测 那么采用GPU机器因为其不支持 mutilprocessing 是不是就是会比CPU机器还要慢的许多？

[littletomatodonkey]

你好，如果希望bs>1的识别预测，可以参考这里进行修改， /tools/infer/predict_rec.py中是支持bs>1的识别预测的

https://github.com/PaddlePaddle/PaddleOCR/blob/488486587b219910e6863bdd1bbd65017fc6de2c/tools/infer/predict_rec.py#L85

[BobDu]

你好，如果希望bs>1的识别预测，可以参考这里进行修改， /tools/infer/predict_rec.py中是支持bs>1的识别预测的

https://github.com/PaddlePaddle/PaddleOCR/blob/488486587b219910e6863bdd1bbd65017fc6de2c/tools/infer/predict_rec.py#L85

非常感谢您的答复 我个人认为在生产化环境serving部署中 支持batch_size 是很有必要的一个特性 请问社区官方有计划增加这一特性吗？ 如果有需要我可以帮助提供PR