Deep learning - Detect Objects - all run fine but not object is detected

[seisn]

Hi 

I am following this tutorial very closely to detect features on a DEM (geo tiff) using deep learning.

https://youtu.be/gJeMCBJwCrA?t=1442

When exporting the training data I am using the CNN Masks metadata as shown in the tutorial. The only difference from the tutorial is that I am creating the model in Arc using Train Deep Learning Model with model type set to MaskRCNN (Obiect detection). I then run Detect Objects Using Deep Learning as shown in the tutorial. All this runs smoothly with no errors, but I have no object detected.

Does anybody know what the issue here might be?

Thanks.

Nic

[PavanYadav]

Hi @seisn, I am not sure if it's the quality of the model or something else. You can review the model quality in the model_metrics.html report. As a test, 1. you could try to set your threshold really low e.g. 0.1 and see if there is any output 2. use one of your training samples as in put.

Hope this helps

A side note: We have an ArcGIS Image Analyst dedicated community and here you might get response sooner on imagery related issues/questions.

Cheers!

Pavan Yadav | Product Engineer - Imagery and AI
Esri | 380 New York | Redlands, 92373 | USA

https://www.linkedin.com/in/pavan-yadav-1846606/