Method __call__

tools/infer_e2e.py:220–402

img_path: str, optional, default=None Path to the directory containing images or the image filename. save_dir: str, optional, default='e2e_results/' Directory to save prediction and visualization results. Defaults to a subfolder in img_path. is_visual

(self,
                 img_path=None,
                 save_dir='e2e_results/',
                 is_visualize=False,
                 img_numpy=None,
                 rec_batch_num=6,
                 crop_infer=False,
                 return_mask=False,
                 **kwargs)

        }

    def __call__(self,
                 img_path=None,
                 save_dir='e2e_results/',
                 is_visualize=False,
                 img_numpy=None,
                 rec_batch_num=6,
                 crop_infer=False,
                 return_mask=False,
                 **kwargs):
        """
        img_path: str, optional, default=None
            Path to the directory containing images or the image filename.
        save_dir: str, optional, default='e2e_results/'
            Directory to save prediction and visualization results. Defaults to a subfolder in img_path.
        is_visualize: bool, optional, default=False
            Visualize the results.
        img_numpy: numpy or list[numpy], optional, default=None
            numpy of an image or List of numpy arrays representing images.
        rec_batch_num: int, optional, default=6
            Batch size for text recognition.
        crop_infer: bool, optional, default=False
            Whether to use crop inference.
        """

        if img_numpy is None and img_path is None:
            raise ValueError('img_path and img_numpy cannot be both None.')
        if img_numpy is not None:
            if not isinstance(img_numpy, list):
                img_numpy = [img_numpy]
            results = []
            time_dicts = []
            for index, img in enumerate(img_numpy):
                ori_img = img.copy()
                if return_mask:
                    dt_boxes, rec_res, time_dict, mask = self.infer_single_image(
                        img_numpy=img,
                        ori_img=ori_img,
                        crop_infer=crop_infer,
                        rec_batch_num=rec_batch_num,
                        return_mask=return_mask,
                        **kwargs)
                else:
                    dt_boxes, rec_res, time_dict = self.infer_single_image(
                        img_numpy=img,
                        ori_img=ori_img,
                        crop_infer=crop_infer,
                        rec_batch_num=rec_batch_num,
                        **kwargs)
                if dt_boxes is None:
                    results.append([])
                    time_dicts.append({})
                    continue
                res = [{
                    'transcription': rec_res[i][0],
                    'points': np.array(dt_boxes[i]).tolist(),
                    'score': rec_res[i][1],
                } for i in range(len(dt_boxes))]
                results.append(res)