Hello, I am looking for a tutorial that can provide with necessary steps to train an image segmentation model.
Hi,
Please check the tutorial below:
<img src="https://github.com/dusty-nv/jetson-inference/raw/master/docs/images/deep-vision-header.jpg" width="100%">
<p align="right"><sup><a href="segnet-pretrained.md">Back</a> | <a href="segnet-patches.md">Next</a> | </sup><a href="../README.md#two-days-to-a-demo-digits"><sup>Contents</sup></a>
<br/>
<sup>Semantic Segmentation</sup></s></p>
# Training FCN-Alexnet with DIGITS
When the previous data import job is complete, return to the DIGITS home screen. Select the `Models` tab and choose to create a new `Segmentation Model` from the drop-down:
<img src="https://github.com/dusty-nv/jetson-inference/raw/master/docs/images/segmentation-digits-create-model.png" width="250">
In the model creation form, select the dataset you previously created. Set `Subtract Mean` to None and the `Base Learning Rate` to `0.0001`. To set the network topology in DIGITS, select the `Custom Network` tab and make sure the `Caffe` sub-tab is selected. Copy/paste the **[FCN-Alexnet prototxt](https://raw.githubusercontent.com/NVIDIA/DIGITS/master/examples/semantic-segmentation/fcn_alexnet.prototxt)** into the text box. Finally, set the `Pretrained Model` to the output that the `net_surgery` generated above: `DIGITS/examples/semantic-segmentation/fcn_alexnet.caffemodel`
Give your aerial model a name and click the `Create` button at the bottom of the page to start the training job. After about 5 epochs, the `Accuracy` plot (in orange) should ramp up and the model becomes usable:
At this point, we can try testing our new model's inference on some example images in DIGITS.
show original
Thanks.
I found a train_segnet.py file in the jetson_inference directory under jetson_inference/python/training/segmentation and I was wondering can this be used to train a custom segmentation model instead of nvidia-digits as digits is not python3 supported and also has been outdated. Thanks
Hi @nitinkumar96 , the training code for the FCN-Resnet18 models in the repo is from here: https://github.com/dusty-nv/pytorch-segmentation/tree/dev
However it is resource intensive to train segmentation networks, and I trained those on a PC with a discrete GPU. There is also a tutorial a community member wrote here: