Hello,
I created a first version (yolov12) of my model with 2000 pictures.
When i test it with a picture belong to the training images, it doesn’t detect objects !
Is it normal ?
I thought it will always detect objects for all trained pictures…
And i created a Roboflow instant model, the results are better.
Any idea ?
Thanks
How are you testing the model? I looked at your project on Universe and the model works well there How to Use the VF_Faucet Object Detection API
Hi,
I tested the model directly in the roboflow interface, on the Universe and with API. Same bad results.
Example:
I’d suggest looking at your dataset splits and class representation to make sure they are balanced VF_Faucet Dataset and Pre-Trained Model by VFFaucet
oh !
you are right ! I have a lot of class without valid and test images !
How can i use “roboflow instant” with API detect.roboflow.com ?