Python OpenCV DNN加载CRNN ONNX模型执行forward时报卷积层错误
OpenCV DNN 加载CRNN OCR模型推理报错
测试基本信息
- 测试目标:基于OpenCV DNN框架运行预训练OCR模型完成文字识别任务
- 识别模型:CRNN_VGG_BiLSTM_CTC.onnx,为OpenCV官方DNN OCR教程配套模型
- 测试输入:样例图片sample-image.png
- 排查诉求:已完成多轮排查,确认该模型理论上支持动态输入形状,优先寻求纯OpenCV框架下的可落地模型加载与推理实现方案
可复现代码
import cv2 as cv import os # 配置识别模型路径 modelRecognition = os.path.join(MODELS_PATH, 'CRNN_VGG_BiLSTM_CTC.onnx') # 加载ONNX格式识别网络 recognizer = cv.dnn.readNetFromONNX(modelRecognition) # 读取测试样例图像 sample_image = cv.imread('sample-image.png') H, W, C = sample_image.shape # 生成适配网络输入的4维Blob blob = cv.dnn.blobFromImage(sample_image, size=(H, W)) recognizer.setInput(blob) # 前向推理执行阶段触发报错 result = recognizer.forward()
触发的报错信息
OpenCV(4.4.0) /tmp/pip-req-build-xgme2194/opencv/modules/dnn/src/layers/convolution_layer.cpp:348: error: (-215:Assertion failed) ngroups > 0 && inpCn % ngroups == 0 && outCn % ngroups == 0 in function 'getMemoryShapes'
内容的提问来源于stack exchange,提问作者Tom
相关产品推荐
相关产品推荐

