使用多语言检测模型转onnx时报错

[yunqianluo]

非常棒的工作。使用多语言检测模型onnx时出现了下面的情况
1、运行环境
numpy 1.26.4
onnxruntime 1.14.1
onnxruntime-gpu 1.14.1
opencv-contrib-python 4.7.0.72
opencv-python 4.11.0.86
opencv-python-headless 4.7.0.72
opt-einsum 3.3.0
packaging 24.2
paddle2onnx 1.2.4
paddlepaddle 2.6.0
2、使用多语言检测模型Multilingual_PP-OCRv3_det_infer出现错误
import cv2
import time
from onnxocr.onnx_paddleocr import ONNXPaddleOcr,sav2Img
import sys

det_model_dir = '/data/app/paddlepaddle/Multilingual_PP-OCRv3_det_infer.onnx'
rec_model_dir = '/data/app/paddlepaddle/cyrillic_PP-OCRv3_rec_infer.onnx'
rec_char_dict_path = '/data/app/paddlepaddle/OnnxOCR-main/onnxocr/dict/cyrillic_dict.txt'
cls_model_dir = '/data/app/paddlepaddle/ch_ppocr_mobile_v2.0_cls_infer.onnx'

det_limit_side_len = 960

model = ONNXPaddleOcr(det_limit_side_len=det_limit_side_len, det_model_dir=det_model_dir, rec_model_dir=rec_model_dir, rec_char_dict_path=rec_char_dict_path, cls_model_dir=cls_model_dir, use_angle_cls=True, use_gpu=False, use_onnx=True)

imgpath = ''
img = cv2.imread('./onnxocr/test_images/ru.png')
s = time.time()
result = model.ocr(img)
e = time.time()
print("total time: {:.3f}".format(e - s))
print("result:", result)
for box in result[0]:
print(box)

sav2Img(img, result)
运行时出现下列错误：

outputs = self.det_onnx_session.run(self.det_output_name, input_feed=input_feed)
File "/home/xxxx/anaconda3paddle/envs/paddle2onnx/lib/python3.10/site-packages/onnxruntime/capi/onnxruntime_inference_collection.py", line 200, in run
return self._sess.run(output_names, input_feed, run_options)
onnxruntime.capi.onnxruntime_pybind11_state.InvalidArgument: [ONNXRuntimeError] : 2 : INVALID_ARGUMENT : Got invalid dimensions for input: x for the following indices
index: 3 Got: 832 Expected: 960
Please fix either the inputs or the model.

请问一下这是什么原因？

[jingsongliujing]

识别模型的输入大小不对，修改这里：https://github.com/jingsongliujing/OnnxOCR/blob/main/onnxocr/utils.py
或者切换到master分支用master分支的代码