It would also be nice to train models for detection and segmentation.
Just ran the SDK on a detection dataset. It is blazing fast! But the accuracy/precision/recall/f1 don't look as good as other models that I've trained on my data. The SDK model is about 5x faster, but the f1 score is 35% lower than the other model I'm looking at.
In the future will users be able to trade-off accuracy with speed? How fine-grained is that trade-off? Are users switching out models or pulling a slider?Reply