Hi
I started working on Jetson Nano. I am unable to work on detectnet model. I executed the model successfully. But I cannot see the appropriate detection on the output video. Unable to find out the errors. Can anyone please help me in that.
Hi,
Could you share more details about your procedure?
How do you inference the detectnet model? Deepstream or jetson_inference?
Thanks.
I used Jetson inference only.
When I used the imagenet model with one short video and saved the result, I got correct results.
But when I use the video for detectnet model and saved the results, I cannot see the results. I have seen some tutorials in net, in that they gave a syntax to execute the model.
The syntax is
./detectnet.py (#path of input video) (#path to save output video)
I used this syntax and gave the appropriate paths for input and output videos. But i cannot see the output video in the directory where I saved the output video. Code is running fine.but no output results. Can you please help me out.
Hi @siddhusiddhartha25, can you post the command line that you ran, along with the console log from running the application?
Can you run the detectnet video file examples from this step of the tutorial?
Sir, I used those tutorials only.
Can you provide the output log from the terminal?
Also are you running it from a directory that you have write permissions to?
Now it is working fine sir.
One more thing, By default, detect net models using 91 labels. Is it possible to add new labels.
You would need to re-train the model, however you would have to re-train it on the previous 91 classes as well.
Instead I recommend only picking the classes that you really want, because the 90 classes from MS COCO is a lot of classes for a detection model. You can follow this tutorial for training your own object detection models:
https://github.com/dusty-nv/jetson-inference/blob/master/docs/pytorch-ssd.md