Open XieKaiwen opened 1 month ago
Can you provide some samples?
Here is an example of the annotated image vs original images
as can be seen, the model is predicting with 1.0 confidence even on random spots.
I do realise the model needs more training, but does this problem get better with more training?
I have tried with my trained model and multiple times the model has detected everything with 1.00 score, even when making mistakes.
Is there anything I am doing wrong, because i think 1.00 score for everything is abit suspicious