I use colab to train my moss dataset, however it can't detect anything in the picture, how do i conquer this problem? To train more pictures? And while labeling the moss with labelimg, i just labeled the moss that i saw in the picture, but there are too many! Is it appropriate to do that? enter image description here
Asked
Active
Viewed 30 times
1 Answers
0
You need to label all the objects in the training data that needs to be detected. If you are not labeling, the model will get confused whether it needs to predict it or not. By not labeling the data, you will be penalising the model when it tries to do a correct prediction.

Ajai
- 1,049
- 1
- 11
- 23