Esri / Esri/raster-deep-learning
Model working under pro but not on image server
- Dominant language
- Python
- Stars
- 203
- Forks
- 92
- PR merge metrics
- No merged PRs in 30d
Description
When using deep learning services on my arcgis image server, I receive this as a response for a QueryDeepLearningModelInfo job
``` json
{
"jobId": "jf07709889e9d4c158427b37e549a6a58",
"jobStatus": "esriJobFailed",
"messages": [
{
"type": "esriJobMessageTypeInformative",
"description": "Submitted."
},
{
"type": "esriJobMessageTypeInformative",
"description": "Executing..."
},
{
"type": "esriJobMessageTypeInformative",
"description": "Start Time: 25 August 2020 10:21:28"
},
{
"type": "esriJobMessageTypeInformative",
"description": "Raster Analytics helper service: https://polaris-analysis.phedre.fr:6443/arcgis"
},
{
"type": "esriJobMessageTypeInformative",
"description": "Running on ArcGIS Image Server."
},
{
"type": "esriJobMessageTypeError",
"description": "'7fdf83ecd5974655b27d5f4bba80c7fa' is not a valid model definition."
},
{
"type": "esriJobMessageTypeError",
"description": "Unable to initialize python raster function with scalar arguments. [Z:\\data\\product\\arcgis\\server\\usr\\directories\\arcgissystem\\resources\\models\\raster\\7fdf83ecd5974655b27d5f4bba80c7fa\\ArcGISImageClassifier.py]"
},
{
"type": "esriJobMessageTypeError",
"description": "Unable to initialize python raster function with scalar arguments."
},
{
"type": "esriJobMessageTypeError",
"description": "Unable to read deep learning model definition."
},
{
"type": "esriJobMessageTypeError",
"description": "Failed to execute (QueryDeepLearningModelInfo)."
},
{
"type": "esriJobMessageTypeInformative",
"description": "Failed at 25 August 2020 10:21:45 (Elapsed Time: 16.94 seconds)"
},
{
"type": "esriJobMessageTypeError",
"description": "Failed."
}
]
}
```
the message below prove that it did find my model where it was installed on the server.
``` json
{
"type": "esriJobMessageTypeError",
"description": "Unable to initialize python raster function with scalar arguments. [Z:\\data\\product\\arcgis\\server\\usr\\directories\\arcgissystem\\resources\\models\\raster\\7fdf83ecd5974655b27d5f4bba80c7fa\\ArcGISImageClassifier.py]"
}
```
but it also tells me my model isn't defined correctly but when I download the files (.emd, ArcGISImageClassifier.py and .pth) and run them using ArcGIS pro "Classify Pixels Using Deep Learning" tool it works flawlessly.
```json
{
"type": "esriJobMessageTypeError",
"description": "'7fdf83ecd5974655b27d5f4bba80c7fa' is not a valid model definition."
}
```
I'm using an Arcgis Image Server 10.8.1 on CentOS without CPU (but did install tensorflow-mkl). I followed this [guide](https://www.esri.com/content/dam/esrisites/en-us/media/pdf/install-deep-learning-frameworks.pdf) to install the required libraries.
joined you can find my model that I tested under Arcgis Pro and the result of `conda list
`
[wine_conda_list.txt](https://github.com/Esri/raster-deep-learning/files/5124377/wine_conda_list.txt)
[model-20.txt](https://github.com/Esri/raster-deep-learning/files/5124396/model-20.txt)
[ArcGISImageClassifier.txt](https://github.com/Esri/raster-deep-learning/files/5124402/ArcGISImageClassifier.txt)
Contributor guide
No contributing guide indexed for this repository
Assessment
This issue has not been assessed yet.