Lesley Natrop

That's okay! Just let me know :)

Hey, want to check again?

Is that something you can fix or is that the Cupra API not working right?

Ah okay! There have been 2 updates today for the app, but they just broke the app. The first update made it so nobody could even open the app. The...

@Laughing-q I'm not sure what you mean with 'negative labels'. There are no empty labels in the dataset. Adding v5loader=True to the query worked, it's now training

@Laughing-q I made my dataset with Roboflow and exported it in YoloV8 format. https://universe.roboflow.com/lesley-natrop-zgywz/traffic-detection-e3og7/dataset/6 This is the dataset. Hope this helps

Thank you for the quick help! That makes sense, I will fix those labels and will try again.

No pass v5loader=True as argument when training.

I'm absolutely no expert in this. When there's not enough vram, the error would be something like: Cuda out of memory

Getting back to this, `hide_labels=True` does indeed hide the label for detection, but I want the bounding box for the segmentation to not be visible. `line_thickness=0` does not work either.