Hi, I’ve been using the Jetson Nano for ML inference and everything is alright except the segnet-camera and segnet-console. I’m using the FCN-AlexNet-Cityscapes-SD as the model. I’ve attached a screenshot of the output I am being provided with by the Jetson Nano.
The input I provided was Input: drone_0255.png - Album on Imgur
And the output I recieved was: Output (segnet-console) - Imgur
I’m a newbie in AI and ML inference. Kindly help me correct the output.
Hi ssummar, the image you are testing on is from the aerial-fpv dataset, but the model is trained on Cityscapes - so you would want to run the Cityscapes model on test images from the Cityscapes dataset or other urban street scenes (like from KITTI dataset or similar scenes).
Also, I’ve been working on newly-trained models for segmentation that have better performance, see here:
[url]https://devtalk.nvidia.com/default/topic/1051696/jetson-nano/how-to-run-schematic-segmentation-samples-in-nano/post/5377005/#5377005[/url]