海思SS928V100，svp_npu部署yolov8自训练模型，离线om模型转换过程中报错

我转的yolov5的.wk，中间问题挺多的，一个一个问题磨才解决的，不容易啊

自己摸索解决了，还得慢慢问ai

你好，请问下怎么解决的

你好，我现在yolov8官网下的模型转换出问题了，方便分析一下你的rpn补丁吗，我各种方法都试过了，就是跑不起来

您好，我转换onnx模型的时候一直报错，总是自动给我用onnx的opset1，然后就说不支持
ONNX: starting export with onnx 1.19.1 opset 13…
[W1023 07:45:00.354568488 shape_type_inference.cpp:1995] Warning: The shape inference of custom_ops::DecBBox type is missing, so it may result in wrong shape inference for the exported graph. Please consider adding it in symbolic function. (function UpdateReliable)
[W1023 07:45:00.370685601 shape_type_inference.cpp:1995] Warning: The shape inference of custom_ops::Filter type is missing, so it may result in wrong shape inference for the exported graph. Please consider adding it in symbolic function. (function UpdateReliable)
[W1023 07:45:00.373652233 shape_type_inference.cpp:1995] Warning: The shape inference of custom_ops::Sort type is missing, so it may result in wrong shape inference for the exported graph. Please consider adding it in symbolic function. (function UpdateReliable)
[W1023 07:45:00.376280895 shape_type_inference.cpp:1995] Warning: The shape inference of custom_ops::NMS type is missing, so it may result in wrong shape inference for the exported graph. Please consider adding it in symbolic function. (function UpdateReliable)
ONNX: export failure ❌ 0.6s: Exporting the operator ‘custom_ops::DecBBox’ to ONNX opset version 1 is not supported. Support for this operator was added in version 12, try exporting with this version
这是我的报错，请问这个该怎么解决

学习下