Classify Pixel Using Deep Learning -- blank image

[forestfairy]

Hi, 

I'm currently working on satellite images (Planet images with 4 bands) and I have to apply deep Learning model to classify them. I exported the training datas with the right size of image (3 bands and unsigned 8-bits pixels) and managed to run the Deep Learning models.

However when I try to apply them on my image, The program says it works but the resulting raster is empty : either the statistics are all at 0 or it doesn't calculate them. 

I join what I did and the results. 

If anyone could explain to me why it doesn't work, I'd be really grateful.

[DanPatterson]

I moved your post to

Imagery and Remote Sensing Questions - Esri Community

where you are more likely to get an interested and informed audience

[forestfairy]

Thanks a lot. I hope to get an answer.

[VinayViswambharan]

Hi Forestfairy - 

Please check out this thread. They had a similar problem. 

https://community.esri.com/t5/imagery-and-remote-sensing-questions/classify-pixels-using-deep-learni...

[forestfairy]

Hi, 

Thank you for your answer. I had troubles with my computer so I couldn't check before. I read the thread and I understood that their problem was that the raster didn't have the correct pixel depth. 

That's not my case because I converted the raster and had this corrected after troubles with the deep learning training. So I exported my training data with an 8 bits unsigned and 3 bands raster but it still doesn't work. I join an example of what I did to export my training data :

Therefore I still don't know what's wrong with my models and why it doesn't work.