nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Train Deep Learning model with Tree Detection Pre-trained Model

SadraDehghan

I'm trying to detect trees for a project from aerial imagery and trying to create a tree detection model that does a close to accurate job (the pretrained tree detection model gets less than half the trees and captures many object that aren't trees). I have set up a small amount of training data (I don't really have time to do much more) and when I train a model with just my training data it does a good job at only identifying trees but misses a lot of them.

Is there a way I could use the pretrained model to finetune the model from the training data I have set up? I tried including the pretrained tree detection model as the advanced option in the "Train Deep Learning Model" tool but I get the following error message:

 File "c:\program files\arcgis\pro\Resources\ArcToolbox\toolboxes\Image Analyst Tools.tbx\TrainDeepLearningModel.tool\tool.script.execute.py", line 327, in execute
    training_model = getattr(arcgis.learn, model_type)
AttributeError: module 'arcgis.learn' has no attribute 'DeepForest'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "c:\program files\arcgis\pro\Resources\ArcToolbox\toolboxes\Image Analyst Tools.tbx\TrainDeepLearningModel.tool\tool.script.execute.py", line 390, in <module>
    execute()
  File "c:\program files\arcgis\pro\Resources\ArcToolbox\toolboxes\Image Analyst Tools.tbx\TrainDeepLearningModel.tool\tool.script.execute.py", line 385, in execute
    del training_model_object
UnboundLocalError: local variable 'training_model_object' referenced before assignment

Thanks!

Find more posts tagged with

Accepted answers

All comments

PavanYadav

seems like you can finetune the tree detection model - https://doc.arcgis.com/en/pretrained-models/latest/imagery/finetuning-the-palm-tree-detection.htm

AndrewBivins

So is there any way to build on these models to advance our tree detection faster? Or do we have to start from scratch?

Thanks,

LukasFalk

Pulled from (https://learn.arcgis.com/en/projects/improve-a-deep-learning-model-with-transfer-learning/).

VaibhavR

Hi all,

Tree detection model : https://arcg.is/0yLem5
Tree segmentation model : https://arcg.is/1OKPS10
Both are 3rd party models and fine-tuning or transfer learning is not support from them.

PavanYadav

@Student123 Can you share the .emd file created by the Export Training Data tool?

Student123

I have the same problem and I dont know how to fix it. I use the Tree Detection Model of Esri (Deep Forest), my ArcGIS Pro Version is 3.2 and I can train my own model from scratch and I can also use this model, but I always get the error when I want to do Transfer Learning with the pretrained Model

PavanYadav

Hello @SadraDehghan. I am assuming you're using the "Pre-trained Model" option available under the advance section. Are you using one of the pre-trained models available on the ArcGIS Living Atlas? Can you please share the name or link to the model? Can you please also share the metadata format used to create your additional training data?

Join the Conversation

Signed in members can post, follow updates, and more. New here? Register a free account.

Quick Links

Community Help
Find useful guides, FAQs, and documents to help you navigate and make the most of Esri Community.