When I run the interactive_regression notebook on the Orin Nano, I don’t get a camera widget that shows the camera feed, instead it just prints VBOX(…)
I have ipywidgets installed. How can I get the camera feed working in Jupyter Notebook?
The Orin Nano has the GUI enabled currently, and while connected to a display I was able to run all the jetson-inference training examples–including using the live camera feed, so I know the camera works
The same camera can successfully run jetson-inference detectnet to identify objects from a live camera feed, so I’m certain the camera is functioning as expected
When I run nvgstcapture-1.0 when connected to a display, I get a display output
The error I’m getting is: “Failed to load model class ‘ClickableImageModel’ from module ‘jupyter_clickable_image_widget’”
How do I resolve this error? When I try to install jupyter_clickable_widgets from the source, it says the requirement is already met (source: GitHub - jaybdub/jupyter_clickable_image_widget)
I used the below command to install both 7.6.5 & 7.7.2 and they completely break the widgets, instead of showing anything it just shows a rectangle javascript error; only the latest ipywidgets work
I need the ClickableImageModel to work in order to take images of the track to build the dataset that’s used to train the model
The issue is jupyter_clickable_image_widget relies on libraries that are deprecated/no longer used so I don’t know how people will be able to use JetRacer on newer devices if that’s the case
jupyter labextension throws an error saying it needs nodejs version >= 18.0.0 but the latest version of nodejs I can install on the orin is v10.19.0–is there a way to get around this?
I’m also unable to manually import ClickableImageModel running the command
$ python3
>>> from jupyter_clickable_image_widget import ClickableImageModel
I no longer have access to the Orin Nano (it was part of our senior project and the semester ended a few days ago so we returned everything back to the school)–that being said, another group will be picking up where my team left off in the Fall semester, so I updated my notes that I provided for the next team pointing to your modified fork to resolve the clickable image widget issue
An updated working Jupyter Clickable Image Widget will be great so the road following model can be trained directly on the Orin Nano, which prior to your update was not working
Venkat
Here’s our team’s project summary if anyone is curious, we were able to get JetRacer, variable throttle, object detection (via jetson-inference and mobilenet ssd v2) and avoidance working on our Orin Nano:
