Hey, I want to train a FasterRCNN Resnet50 FPN with Pytorch and get weird COCO Eval values. I'm comparing training with real and synthetic images. Please help me, it's my bachelor thesis and I'm struggeling af
#[NEED HELP] Computer Vision problem with Model training
ancient turret
rugged pulsar
post the notebook or whatever code you have so far
heavy nymph
yes
rugged pulsar
so what are the weird values?
ancient turret
I managed to fix it already.. thanks anyway!
rugged pulsar
I managed to fix it already.. thanks anyway!
nice, what was the issue after all?
ancient turret
half of my data had bounding boxes as float and the other integer, looks like my data was causing the problem
rugged pulsar
half of my data had bounding boxes as float and the other integer, looks like my...
ah, good job catching it
python is all fun and games until the type errors start pouring in