Hello everyone.
Is there any way in python to see the SegNet network architecture?
I looked at the jetson documentation but couldn’t find an answer.
Thank you.
Hi @forflafor, do you specifically mean the SegNet DNN architecture from [1511.00561] SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation? If so, you could install PyTorch/TensorFlow/ect and try running it in it’s native framework.
There are some TensorRT-accelerated FCNs in jetson-inference:
Hi @dusty_nv, thanks for the answer.
I was referring to the segnet.py network implemented in the Jetson API.
In Pytorch there is the possibility to print the structure of the network and in fact I wanted to ask how this could be done with the Segnet network.
Since there are, I wanted to know if the FCN-ResNet worked both as an encoder and as a decoder in the SegNet architecture
Thanks.
FCN-ResNet has encoder/decoder and basically functions the same as SegNet, I use it because it came in torchvision, exported to ONNX, and had good runtime performance.
Thanks for the reply @dusty_nv .
As for the number of layers present in FCN-ResNet18, do I take into consideration those of the original paper (https://arxiv.org/pdf/1512.03385.pdf)(which are 18) or have you created a custom architecture?
Thank you.
It’s the 18 layers from ResNet18 plus additional FCN layers.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.