topic Re: Error when training deep learning model in ArcGIS Image Analyst Questions

Error when training deep learning model

IvanVanchugov — Sat, 25 Nov 2023 10:02:19 GMT

I'm trying to train my model and I'm getting the following error:

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 334, in execute
    training_model_object.fit(
  File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\arcgis\learn\models\_arcgis_model.py", line 902, in fit
    lr = self.lr_find(allow_plot=False)
  File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\arcgis\learn\models\_arcgis_model.py", line 721, in lr_find
    raise e
  File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\arcgis\learn\models\_arcgis_model.py", line 718, in lr_find
    self.learn.lr_find()
  File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\fastai\train.py", line 41, in lr_find
    learn.fit(epochs, start_lr, callbacks=[cb], wd=wd)
  File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\fastai\basic_train.py", line 200, in fit
    fit(epochs, self, metrics=self.metrics, callbacks=self.callbacks+callbacks)
  File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\fastai\basic_train.py", line 99, in fit
    for xb,yb in progress_bar(learn.data.train_dl, parent=pbar):
  File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\fastprogress\fastprogress.py", line 47, in __iter__
    raise e
  File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\fastprogress\fastprogress.py", line 41, in __iter__
    for i,o in enumerate(self.gen):
  File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\fastai\basic_data.py", line 75, in __iter__
    for b in self.dl: yield self.proc_batch(b)
  File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\torch\utils\data\dataloader.py", line 517, in __next__
    data = self._next_data()
  File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\torch\utils\data\dataloader.py", line 557, in _next_data
    data = self._dataset_fetcher.fetch(index)  # may raise StopIteration
  File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\torch\utils\data\_utils\fetch.py", line 44, in fetch
    data = [self.dataset[idx] for idx in possibly_batched_index]
  File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\torch\utils\data\_utils\fetch.py", line 44, in <listcomp>
    data = [self.dataset[idx] for idx in possibly_batched_index]
  File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\fastai\data_block.py", line 657, in __getitem__
    if self.item is None: x,y = self.x[idxs],self.y[idxs]
  File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\fastai\data_block.py", line 120, in __getitem__
    if isinstance(idxs, Integral): return self.get(idxs)
  File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\fastai\vision\data.py", line 271, in get
    res = self.open(fn)
  File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\arcgis\learn\models\_maskrcnn_utils.py", line 144, in open
    self.index_dir[self.inverse_class_mapping[fn[k].parent.name]]
KeyError: 'Class_0'

Failed script (null)...

Failed to execute (TrainDeepLearningModel).

Here are the parameters I'm using to train the model on a sample area:

I have spent quite a lot of time, maybe you will have ideas how to fix it.

Perhaps someone has already encountered something similar.

Re: Error when training deep learning model

PavanYadav — Mon, 27 Nov 2023 17:04:28 GMT

@IvanVanchugov
I understand you're using MaskRCNN model type. What is the metadata type of your training data?

This may help https://pro.arcgis.com/en/pro-app/latest/tool-reference/image-analyst/overview-of-the-deep-learning-models.htm

Thanks

Pavan

Re: Error when training deep learning model

PriyankaTuteja — Tue, 28 Nov 2023 02:47:56 GMT

@IvanVanchugov Please provide the following details for our team to better understand the issue:

1. ArcGIS Pro Version

2. A small sample of your training data for reproducing issue

3. What is class_0 in your data? Is it the name of a class?

Re: Error when training deep learning model

IvanVanchugov — Tue, 28 Nov 2023 08:39:50 GMT

Yes, I'm using MaskRCNN model type and I exported all my data on RCNN masks.

I tried to follow the instructions from this video:

Deep Learning Object Detection Workflow in ArcGIS Pro - YouTube

However, it did not say about possible errors and how to solve them(

Re: Error when training deep learning model

IvanVanchugov — Tue, 28 Nov 2023 08:45:12 GMT

1) Yes (version 3.0.2);

2) We have more then 70 classes of data (I wrote about the video above, it's not clear if this is enough or not)

3) Yes, it was the name of a class, but I deleted it and now I have this:

I also solved problems with installing CUDA and so on to be sure that I would be able to run this training module:

Re: Error when training deep learning model

PriyankaTuteja — Tue, 28 Nov 2023 09:56:06 GMT

Where have you deleted this class_0 from that it made the tool run successfully?

Re: Error when training deep learning model

IvanVanchugov — Fri, 01 Dec 2023 13:32:36 GMT

Yes, thank you for your help!

It did it successfully!)

But, after 3 days, when it finally completed I have only 6 detected objects...

Maybe I don't have enough data to train the model (there were about 80 of them). I have an assumption that it is necessary to give about 1000 classes for training.

This is my model_metrics file:

Re: Error when training deep learning model

PavanYadav — Thu, 04 Jan 2024 00:27:25 GMT

@IvanVanchugov I wrote the following two blogs on training data creation; you might find them helpful:

Tips for labeling images for object detection models

Tips for training data preparation for object detection models

I understand you're using maskrcnn but some of the tips might be still useful.

Thanks

Pavan