topic Re: Train Deep Learning Model Zero Division Error in ArcGIS Image Analyst Questions

Train Deep Learning Model Zero Division Error

Ed_ — Tue, 04 Feb 2025 21:14:48 GMT

The Train Deep Learning Model is failing on me in ArcGIS Pro 3.4.2 even though I have even number of training features.

I am btw also using a pre-trained model (Water Body Extraction (SAR) - USA) as well in the tool.

Validation set at 10

Here's a snippet of stat file

Tool

Error

@PavanYadav any clues?

Re: Train Deep Learning Model Zero Division Error

Ed_ — Wed, 05 Feb 2025 17:55:31 GMT

@BrendanNewell @ShivaniPathak can you please help with this? Thank you!

Re: Train Deep Learning Model Zero Division Error

ShivaniPathak — Wed, 12 Feb 2025 05:19:48 GMT

Hi @Ed_ , could you please confirm if you've referred to the guide for creating the training data?

Re: Train Deep Learning Model Zero Division Error

ShivaniPathak — Wed, 12 Feb 2025 05:36:23 GMT

Also which ArcGIS Pro version are you using?

Re: Train Deep Learning Model Zero Division Error

Ed_ — Wed, 12 Feb 2025 17:38:51 GMT

Hi Shivani, thank you for the reply, yes I basically created a very small training subset from the USA Wetlands data available on the Living Atlas. In that subset I then added an attribute for class value of 1.

Re: Train Deep Learning Model Zero Division Error

Ed_ — Wed, 12 Feb 2025 20:47:36 GMT

I am on the latest version of Pro that is 3.4.2. I am surprised that it's giving me the zero division error, because as you can see in the above stat file screenshot, the training data has 600 features.

Re: Train Deep Learning Model Zero Division Error

PavanYadav — Fri, 14 Feb 2025 19:40:37 GMT

hi @Ed_ we need to handle this error bit better and provide you with actionable information. One reason that I have seen this occur is that when the batch size is larger than the number of chips in the data, the Train Deep Learning Model tool may return "ZeroDivisionError: division by zero" in this case.

It looks like you have 5 images chips and your batch size if 8.

Re: Train Deep Learning Model Zero Division Error

Ed_ — Fri, 14 Feb 2025 19:47:59 GMT

Hi Pavan, happy Friday, thank you for the response, so in this case what should be batch size then?

Re: Train Deep Learning Model Zero Division Error

PavanYadav — Fri, 14 Feb 2025 19:51:54 GMT

@Ed_ you can try 2, but to be honest; I would have at least 50-100 samples just to be on the safer side.. I see your tile size is 512, which could be fairly large for your GPU.. maybe decease it to 256 and use a stride of 128.. these will produce a lot more image chips..

Re: Train Deep Learning Model Zero Division Error

Ed_ — Fri, 14 Feb 2025 20:24:05 GMT

i do have 600 training features, so what do you mean by 50-100 samples? Are those the number of image chips?

I do have Nvidia RTX A2000 12 gig, so should i still decrease the tile size?

Re: Train Deep Learning Model Zero Division Error

Ed_ — Tue, 18 Feb 2025 18:43:55 GMT

@PavanYadav ?

Re: Train Deep Learning Model Zero Division Error

PavanYadav — Tue, 11 Mar 2025 23:39:32 GMT

@Ed_ looks like you have 6 image chips in the training data..