How to prepare a training dataset for object detection?

Hi,

I will train a model to detect some objects based on aerial photos.

For this I used these instruvtions decribed here:

https://engage.safe.com/blog/2019/01/fme-does-computer-vision/

After indicating the objets witch OpenCV I will preparte the data for training witch " RasterObjectDetectorSamplePreparer". Inputs are the source directory of the negative files and the annotation file.

If I execute this I get an error:

I don´t know what the problem !

Does anyone have experience in this regard?