search

dusty-nv / pytorch-segmentation

Training of semantic segmentation networks with PyTorch
34 stars 39 forks source link

Quick tutorial for custom datasets? #3

Closed marvision-ai closed 4 years ago

marvision-ai commented 4 years ago

Hi @dusty-nv ,

Thanks for all the work you do on the inference repo. I realized you are using this repo to provide the models for the segmentation.

I was wondering if there was a quick and simple tutorial somewhere that explains the best way to use this repo in order to train on custom datasets and then use with the jetsoninfernce libs.

Thanks!

dusty-nv commented 4 years ago

Hi mbufi, this was forked from the torchvision examples: https://github.com/pytorch/vision/blob/master/references/segmentation/README.md

I added support for fcn-resnet18, fcn-resnet34, and fcn-resnet50, and added support for a bunch of datasets. I do plan to make a tutorial around this for Hello AI World, hopefully in not too long. I need to finish the one for object detection first (pytorch-ssd).

Note that if you want to use this, see my fork of torchvision, the 0.3.0 branch. It is there that there is fcn-resnet18, ect. I need to move these into this repo from torchvision fork before deploying the tutorial.

From: mbufi notifications@github.com Sent: Monday, June 15, 2020 1:30:01 PM To: dusty-nv/pytorch-segmentation pytorch-segmentation@noreply.github.com Cc: Dustin Franklin dustinf@nvidia.com; Mention mention@noreply.github.com Subject: [dusty-nv/pytorch-segmentation] Quick tutorial (#3)

Hi @dusty-nvhttps://github.com/dusty-nv ,

Thanks for all the work you do on the inference repo. I realized you are using this repo to provide the models for the segmentation.

I was wondering if there was a quick and simple tutorial somewhere that explains the best way to use this repo in order to train on custom datasets and then use with the jetsoninfernce libs.

Thanks!

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/dusty-nv/pytorch-segmentation/issues/3, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ADVEGK5J2MFPVBBY6ZJABJLRWZLBTANCNFSM4N6MIYXQ.

marvision-ai commented 4 years ago

@dusty-nv Thats great news! Thank you for that. I would love to utilize more of the repo you have been working on.

I am excited to see what you bring. Cheers

donkorechi commented 2 years ago

Hi @dusty-nv,

Is the possibility of quick tutorial for custom dataset still in consideration? I loved the tutorial that was made for Obstacle Detection. It would be nice if the same is done for segmentation. Thanks

dusty-nv commented 2 years ago

Hi @donkorechi, for now I would refer you to this tutorial on training segmentation models that work with jetson-inference library: https://www.highvoltagecode.com/post/edge-ai-semantic-segmentation-on-nvidia-jetson