PaddlePaddle/PaddleOCR · error · ValueError

box_type can only be one of ['quad', 'poly']

Error message

box_type can only be one of ['quad', 'poly']

What it means

ValueError from DBPostProcess.__call__ during inference/postprocess when self.box_type is neither 'quad' nor 'poly'. The constructor did not validate box_type, so an invalid value survives until the first image is decoded and each result row is converted, then the else-branch raises.

Source

Thrown at ppocr/postprocess/db_postprocess.py:253

        for batch_index in range(pred.shape[0]):
            src_h, src_w, ratio_h, ratio_w = shape_list[batch_index]
            if self.dilation_kernel is not None:
                mask = cv2.dilate(
                    np.array(segmentation[batch_index]).astype(np.uint8),
                    self.dilation_kernel,
                )
            else:
                mask = segmentation[batch_index]
            if self.box_type == "poly":
                boxes, scores = self.polygons_from_bitmap(
                    pred[batch_index], mask, src_w, src_h
                )
            elif self.box_type == "quad":
                boxes, scores = self.boxes_from_bitmap(
                    pred[batch_index], mask, src_w, src_h
                )
            else:
                raise ValueError("box_type can only be one of ['quad', 'poly']")

            boxes_batch.append({"points": boxes})
        return boxes_batch


class DistillationDBPostProcess(object):
    def __init__(
        self,
        model_name=["student"],
        key=None,
        thresh=0.3,
        box_thresh=0.6,
        max_candidates=1000,
        unclip_ratio=1.5,
        use_dilation=False,
        score_mode="fast",
        box_type="quad",
        **kwargs,

View on GitHub (pinned to 2661c7c0ef)

Solutions

1. Set PostProcess.box_type to 'quad' or 'poly' in the detection config.
2. If you need quadrilateral output for legacy label format, use 'quad'; for full polygon/NMS-free output use 'poly'.
3. Validate the config before launching a long evaluation run, since the error only surfaces at the first postprocessed batch.

Example fix

# before
PostProcess:
  name: DBPostProcess
  box_type: polygon

# after
PostProcess:
  name: DBPostProcess
  box_type: poly

BOX_TYPES = {"quad", "poly"}
box_type = cfg['PostProcess']['box_type'] if 'PostProcess' in cfg else 'quad'
assert box_type in BOX_TYPES, f"PostProcess.box_type must be one of {BOX_TYPES}, got {box_type!r}"

def is_valid_box_type(v) -> bool:
    return v in ("quad", "poly")

When it happens

Trigger: Running DB/DB++ text detection inference or evaluation with PostProcess.box_type set to anything other than 'quad' or 'poly' in the detection config (e.g. 'polygon', 'box', 'rect').

Common situations: Hand-edited det configs (det_mv3_db.yml etc.) where users write the long form 'polygon' instead of 'poly'; configs copied from other repos (e.g. MMOCR vocabulary) with different box-type names.

Related errors

• not support limit type, image
• Detection model session is not initialized.
• Unexpected det output dims: [${dims.join(", ")}]
• Unexpected det output dims: [${od.join(", ")}]
• Detection batch output N=${String(nOut)} does not match inpu

AI-assisted analysis of PaddlePaddle/PaddleOCR@2661c7c0ef (2026-08-14). Data as JSON: /api/errors/acf2eb7e1dd73e3e. Report an issue: GitHub.