onnx转换失败

[1124676457]

我发现使用onnx转换后，输出是11个，不是代码里面的三个，而且我用shape 768 1344去转换，直接报错
Traceback (most recent call last):
File "onnx_exporter.py", line 231, in
check(args, dummy_input, check_onnx=True, check_trt=False)
File "onnx_exporter.py", line 126, in check
sess = rt.InferenceSession(args.out)
File "/home/tao/anaconda3/lib/python3.8/site-packages/onnxruntime/capi/session.py", line 195, in init
self._create_inference_session(providers, provider_options)
File "/home/tao/anaconda3/lib/python3.8/site-packages/onnxruntime/capi/session.py", line 205, in _create_inference_session
sess.initialize_session(providers or [], provider_options or [])
onnxruntime.capi.onnxruntime_pybind11_state.Fail: [ONNXRuntimeError] : 1 : FAIL : Node (Concat_637) Op (Concat) [ShapeInferenceError] Can't merge shape info. Both source and target dimension have values but they differ. Source=49 Target=48 Dimension=2

这个是我在single_stage_ins 加了你介绍的代码后出现的，不加就不会报错，但输出就变成10个了，求教

[1124676457]

@chenjianqu

[chenjianqu]

输出11个张量是对的, 因为Solov2采用了FPN架构.所有输出的张量维度为: {1, 128, 12, 12},
{1, 128, 16, 16},
{1, 128, 24, 24},
{1, 128, 36, 36},
{1, 128, 40, 40},
{1, 80, 12, 12},
{1, 80, 16, 16},
{1, 80, 24, 24},
{1, 80, 36, 36},
{1, 80, 40, 40},
{1, 128, 96, 288},
我的代码里也是这样定义的