Jetson Nano Jetbot Install "create-sdcard-image-from-scratch" pytorch vision error

[RobotRoss]

Error at :
git clone GitHub - pytorch/vision: Datasets, Transforms and Models specific to Computer Vision
cd vision
#git checkout v0.4.0
sudo -H python3 setup.py install

/home/jetbot/vision/torchvision/csrc/ops/autocast/deform_conv2d_kernel.cpp: In function ‘ at::Tensor vision::ops::
{anonymous}::deform_conv2d_autocast(const at::Tensor&, const at::Tensor&, const at::Tensor&, const at::Tensor&, const at::Tensor&, int64_t, int64_t, int64_t, int64_t, int64_t, int64_t, int64_t, int64_t, bool) ’:
/home/jetbot/vision/torchvision/csrc/ops/autocast/deform_conv2d_kernel.cpp:11:12:

warning: ‘ at::Tensor vision::ops::{anonymous}::deform_conv2d_autocast(const at::Tensor&, const at::Tensor&, const at::Tensor&, const at::Tensor&, const at::Tensor&, int64_t, int64_t, int64_t, int64_t, int64_t, int64_t, int64_t, int64_t, bool) ’ defined but not used [ -Wunused-function ]

at::Tensor deform_conv2d_autocast (
^~~~~~~~~~~~~~~~~~~~~~
error: command ‘aarch64-linux-gnu-gcc’ failed with exit status 1
jetbot@jetbot-ROS2 : ~/jetbot/scripts $

Thus unable to continue installing Jetbot code, i/e/ torch2RT, tartlets, Jupiter, etc., JetBot services, dependencies

.

[jaybdub]

Hi Phreddog24,

Thanks for reaching out!

This may be the result of with a mismatch between the version of PyTorch and Torchvision. For using JetBot, the latest validated instructions are located here

https://jetbot.org/v0.4.3/software_setup/sd_card.html

If you’re interested in doing a custom setup and not using these docker containers, I would recommend following this post for installing PyTorch on Jetson

PyTorch for Jetson

Below are pre-built PyTorch pip wheel installers for Jetson Nano, TX1/TX2, Xavier, and Orin with JetPack 4.2 and newer. Download one of the PyTorch binaries from below for your version of JetPack, and see the installation instructions to run on your Jetson. These pip wheels are built for ARM aarch64 architecture, so run these commands on your Jetson (not on a host PC). You can also use the pre-built l4t-pytorch and l4t-ml container images and Dockerfiles. PyTorch pip wheels JetPack 5

PyTo…

Please let me know if this helps, or if you run into any questions / issues.

Best,
John

[RobotRoss]

Yes, I tried that image on a fresh 64GB SD first which finished ok and JB runs properly but encountered an odd problem regarding displayed SD Card allocations. df -h shows 22 of 24 GB used & several other smaller allocations totaling 7GB, so where is the remaining 35 GB? I tried a Linux command to expand to fill the SD which stated “nothing to do all 64 GB allocated”. I want to also add a ROS2 Foxy Docker to try an additional Jetbot_ros2 project but concerned there will not be sufficient space.

[RobotRoss]

I was able resolve my issue of correctly configuring a fully configured Jetbot 4gb jp45 on a 64gb sd card, with a patch to the Jetbot flash described in another Post. Thus not necessary for me to build Pytorch & torch vision from source. I can now complete my intended installation of a ROS2 Foxy Container on a Jetbot Nano. There may be others who may need to manually configure a Jetson Nano “from scratch”.