Question

• 如何将产出层的层面纳入内心网?

我可以查阅图表,但产出方面没有任何方面:

~/onnx-tensorrt/third_party/onnx/onnx/tools/net_drawer.py --input ./weights/tiny_3l_v5_11_608.onnx  --output ./weights/tiny_3l_v5_11_608.dot --embed_docstring

Terveisin, Markus

Answer 1

你们只能使用

from onnx import shape_inference
inferred_model = shape_inference.infer_shapes(original_model)

并在<条码>中找到星体。

For full doc, see https://github.com/onnx/onnx/blob/master/docs/ShapeInference.md

Answer 2

Hei,

https://github.com/onnx/onnx-tensorrt has this feature.

我这样做了。

onnx2trt yolo_tiny_3l.onnx  -t yolo_tiny_3l.onnx.txt -l -v

Terveisin, Markus

Answer 3

If you are using Java apis to load ONNX model, then get the metadata information from session. Metadata contains the details about input and output dimensions.

Answer 4

Tryloading the .onnx file into this online ON Romudioiser tool (Netron)

Answer 5

You can also use model.graph.output to print the output shape:

from onnx import shape_inference
inferred_model = shape_inference.infer_shapes(original_model)
print(inferred_model.graph.output)

友情链接