运行姿态检测代码遇RET_CHECK失败问题求助（已装依赖）

问题描述 #

运行基于cvzone和mediapipe的姿态检测代码时触发RET_CHECK failure错误，核心提示为Output tensor range is required，已安装相关依赖但问题未解决。

运行代码 #

import cv2
from cvzone.PoseModule import PoseDetector

cap = cv2.VideoCapture('Video.mp4')

detector = PoseDetector()
while True:
    success, img = cap.read()
    img = detector.findPose(img)
    lmList, bboxInfo = detector.findPosition(img)
    cv2.imshow("Image", img)
    cv2.waitKey(1)

完整报错栈 #

Traceback (most recent call last):
  File "/Users/chashanefaith.babanto/Desktop/CRYSTHEL/Motion Capture/MotionCap.py", line 6, in <module>
    detector = PoseDetector()
  File "/Users/chashanefaith.babanto/Desktop/CRYSTHEL/Motion Capture/.venv/lib/python3.9/site-packages/cvzone/PoseModule.py", line 42, in __init__
    self.pose = self.mpPose.Pose(static_image_mode=self.staticMode,
  File "/Users/chashanefaith.babanto/Desktop/CRYSTHEL/Motion Capture/.venv/lib/python3.9/site-packages/mediapipe/python/solutions/pose.py", line 146, in __init__
    super().__init__(
  File "/Users/chashanefaith.babanto/Desktop/CRYSTHEL/Motion Capture/.venv/lib/python3.9/site-packages/mediapipe/python/solution_base.py", line 248, in __init__
    self._graph = calculator_graph.CalculatorGraph(
RuntimeError: ValidatedGraphConfig Initialization failed.
ImageToTensorCalculator: ; RET_CHECK failure (mediapipe/calculators/tensor/image_to_tensor_calculator.cc:144) ValidateOptionOutputDims(options) returned INTERNAL: ; RET_CHECK failure (./mediapipe/calculators/tensor/image_to_tensor_utils.h:136) options.has_output_tensor_float_range() || options.has_output_tensor_int_range() || options.has_output_tensor_uint_range()Output tensor range is required. 
InferenceCalculatorCpu: ; RET_CHECK failure (mediapipe/calculators/tensor/inference_calculator_cpu.cc:63) !options.model_path().empty() ^ kSideInModel(cc).IsConnected()Either model as side packet or model path in options is required.
SplitDetectionVectorCalculator: The number of output streams should match the number of ranges specified in the CalculatorOptions.
ImageToTensorCalculator: ; RET_CHECK failure (mediapipe/calculators/tensor/image_to_tensor_calculator.cc:144) ValidateOptionOutputDims(options) returned INTERNAL: ; RET_CHECK failure (./mediapipe/calculators/tensor/image_to_tensor_utils.h:136) options.has_output_tensor_float_range() || options.has_output_tensor_int_range() || options.has_output_tensor_uint_range()Output tensor range is required. 
ConstantSidePacketCalculator: ; RET_CHECK failure (mediapipe/calculators/core/constant_side_packet_calculator.cc:64) (cc->OutputSidePackets().NumEntries(kPacketTag))==(options.packet_size())Number of output side packets has to be same as number of packets configured in options.
ConstantSidePacketCalculator: ; RET_CHECK failure (mediapipe/calculators/core/constant_side_packet_calculator.cc:64) (cc->OutputSidePackets().NumEntries(kPacketTag))==(options.packet_size())Number of output side packets has to be same as number of packets configured in options.
ConstantSidePacketCalculator: ; RET_CHECK failure (mediapipe/calculators/core/constant_side_packet_calculator.cc:64) (cc->OutputSidePackets().NumEntries(kPacketTag))==(options.packet_size())Number of output side packets has to be same as number of packets configured in options.
SplitTensorVectorCalculator: The number of output streams should match the number of ranges specified in the CalculatorOptions.
SplitNormalizedLandmarkListCalculator: The number of output streams should match the number of ranges specified in the CalculatorOptions.
SplitLandmarkListCalculator: The number of output streams should match the number of ranges specified in the CalculatorOptions.

问题原因与解决方法 #

核心原因 #

该错误由mediapipe与cvzone版本不兼容导致：新版mediapipe对张量处理的配置参数增加了严格校验，但旧版cvzone的PoseModule未适配这些新要求，导致Pose模型初始化时配置错误。

解决步骤 #

1. 降级mediapipe到兼容版本
   卸载当前版本，安装适配性较好的0.9.0.1版本：

   pip uninstall mediapipe -y
   pip install mediapipe==0.9.0.1
   

2. 验证cvzone版本（可选）
   若降级mediapipe后仍有问题，可安装稳定版cvzone（如1.5.6）：

   pip install cvzone==1.5.6
   

3. 显式初始化PoseDetector参数（可选）
   避免默认参数引发的适配问题，显式传入参数初始化：

   detector = PoseDetector(staticMode=False, modelComplexity=1, smoothLandmarks=True)
   

4. 检查视频文件路径
   确保Video.mp4在代码运行目录下，或使用绝对路径，避免视频读取失败引发后续错误。

内容的提问来源于stack exchange，提问作者Faith