Function ocr_recognition

ddddocr/api/app.py:336–404

OCR文字识别 - 接收Base64编码的图片

(
    request: OCRRequest,
    background_tasks: BackgroundTasks,
    ocr: bool = Query(default_ocr, description="是否启用OCR功能"),
    det: bool = Query(default_det, description="是否启用目标检测功能"),
    old: bool = Query(default_old, description="是否使用旧版OCR模型"),
    beta: bool = Query(default_beta, description="是否使用Beta版OCR模型"),
    use_gpu: bool = Query(default_use_gpu, description="是否使用GPU加速"),
    device_id: int = Query(default_device_id, description="GPU设备ID"),
    show_ad: bool = Query(default_show_ad, description="是否显示广告")
)

# OCR识别端点 - JSON请求
@app.post("/ocr", response_model=OCRResponse)
async def ocr_recognition(
    request: OCRRequest,
    background_tasks: BackgroundTasks,
    ocr: bool = Query(default_ocr, description="是否启用OCR功能"),
    det: bool = Query(default_det, description="是否启用目标检测功能"),
    old: bool = Query(default_old, description="是否使用旧版OCR模型"),
    beta: bool = Query(default_beta, description="是否使用Beta版OCR模型"),
    use_gpu: bool = Query(default_use_gpu, description="是否使用GPU加速"),
    device_id: int = Query(default_device_id, description="GPU设备ID"),
    show_ad: bool = Query(default_show_ad, description="是否显示广告")
):
    """
    OCR文字识别 - 接收Base64编码的图片
    """
    image = None
    try:
        img_data = _decode_base64_bytes(request.image)
        image = Image.open(io.BytesIO(img_data))
    except HTTPException:
        raise
    except Exception as exc:
        logger.error(f"OCR请求图片解析失败: {str(exc)}")
        raise HTTPException(status_code=400, detail="无法读取图片") from exc

    config_key = f"ocr={ocr}-det={det}-old={old}-beta={beta}-gpu={use_gpu}-dev={device_id}"

    ocr_instance = get_ocr_instance(
        config_key, ocr, det, old, beta, use_gpu, device_id, show_ad,
        default_import_onnx_path, default_charsets_path
    )

    start_time = time.time()
    try:
        if request.probability:
            result = ocr_instance.classification(
                image,
                probability=True,
                colors=request.colors,
                custom_color_ranges=request.custom_color_ranges
            )
            response_data = {
                "result": result,
                "processing_time": time.time() - start_time
            }
        else:
            result = ocr_instance.classification(
                image,
                colors=request.colors,
                custom_color_ranges=request.custom_color_ranges
            )
            response_data = {
                "result": result,
                "processing_time": time.time() - start_time
            }
    except (DdddOcrInputError, InvalidImageError) as exc:
        logger.warning(f"OCR识别参数错误: {str(exc)}")
        raise HTTPException(status_code=400, detail=str(exc)) from exc
    except Exception as exc: