Yeah, this wasn’t supposed to happen. I trained an Instance segmentation model using NVIDIA’s Transfer Learning Toolkit to detect cars and deployed it on a Jetson. It seems I messed up with either the settings for the threshold or with the whole training process. Now every car is like in a star mode in Mario Kart.
Another weird thing, the Jetson NANO was able to at least display the segmentation but the Jetson XAVIER wouldn’t display it at all. I took the same method but it wasn’t working. I haven’t figured it out yet. The GPU’s architecture for each Jetson is different so that might be cause…?
