Detect Object Using Deep Learning Error 999999 When GPU in used

1006
4
Jump to solution
08-14-2021 12:50 AM
bhariadi
New Contributor III

Hello All,

I successfully detect object using deep learning, processed using CPU as followed pictured. 😊

Capture1.PNG

However, when I use GPU to run this tool, the process end with error. 🤔

Capture2.PNGCapture3.png

Those are my laptop specification:

  • DELL Precision 5550
  • Intel(R) Core(TM) i9-10885H CPU @ 2.40GHz 2.40 GHz
  • 32.0 GB of RAM
  • Windows 10 Pro
  • NVIDIA Quadro T2000 with Max-Q Design
  • ArcGIS Pro 2.8.2

I installed ArcGIS Pro Deep Learning Framework, and get the installer from this site.

This is the error messages, I copied from ArcGIS Pro:

Detect Objects Using Deep Learning
=====================
Parameters

Input Raster 47_res10cm_micasensemx.tif
Output Detected Objects C:\Users\Administrator\Documents\ArcGIS\Projects\20210813_ArcGISPro_BelajarDeepLearning_1\20210813_ArcGISPro_BelajarDeepLearning_1.gdb\c47_res10cm_micasensemx_DetectObject1_GPU
Model Definition C:\DeepLearning\models\OPDetectionModel_Res10cm_YOLOv3_v11_Stage1_scorexxxx.dlpk
Arguments padding 56;threshold 0,1;nms_overlap 0,5;batch_size 4;exclude_pad_detections True
Non Maximum Suppression NMS
Confidence Score Field Confidence
Class Value Field Class
Max Overlap Ratio 0,5
Processing Mode PROCESS_AS_MOSAICKED_IMAGE
=====================
Environments

Scratch Workspace C:\Users\Administrator\Documents\ArcGIS\Projects\20210813_ArcGISPro_BelajarDeepLearning_1\20210813_ArcGISPro_BelajarDeepLearning_1.gdb
Extent 288571,076325618 9786953,0828514 288891,75196697 9787250,26344576
Processor Type GPU
Current Workspace C:\Users\Administrator\Documents\ArcGIS\Projects\20210813_ArcGISPro_BelajarDeepLearning_1\20210813_ArcGISPro_BelajarDeepLearning_1.gdb
=====================
Messages

Start Time: 13 August 2021 17:31:38
ERROR 999999: Something unexpected caused the tool to fail. Contact Esri Technical Support (http://esriurl.com/support) to Report a Bug, and refer to the error help for potential solutions or workarounds.
Python raster function is unable to vectorize the data. [Failed to generate table]
Python raster function is unable to vectorize the data.
Traceback (most recent call last):
File "C:\Users\ADMINI~1\AppData\Local\Temp\ArcGISProTemp12504\OPDetectionModel_Res10cm_YOLOv3_v11_Stage1_scorexxxx.dlpk\ArcGISObjectDetector.py", line 215, in vectorize
polygon_list, scores, classes = self.child_object_detector.vectorize(**pixelBlocks)
File "C:\Users\Administrator\AppData\Local\Programs\ArcGIS\Pro\bin\Python\envs\deeplearning282\Lib\site-packages\arcgis\learn\models\_inferencing\_yolov3_inference.py", line 334, in vectorize
for feature_idx in range(len(image_bbox.data[0])):
File "C:\Users\Administrator\AppData\Local\Programs\ArcGIS\Pro\bin\Python\envs\deeplearning282\Lib\site-packages\fastai\vision\image.py", line 378, in data
lbls = np.array([o.data for o in lbls]) if lbls is not None else None
File "C:\Users\Administrator\AppData\Local\Programs\ArcGIS\Pro\bin\Python\envs\deeplearning282\Lib\site-packages\torch\tensor.py", line 486, in __array__
return self.numpy()
TypeError: can't convert CUDA tensor to numpy. Use Tensor.cpu() to copy the tensor to host memory first.
Python Raster Function's .vectorize() method returned nothing.
Python raster function is unable to vectorize the data. [Failed to generate table]
Python raster function is unable to vectorize the data.
Traceback (most recent call last):
File "C:\Users\ADMINI~1\AppData\Local\Temp\ArcGISProTemp12504\OPDetectionModel_Res10cm_YOLOv3_v11_Stage1_scorexxxx.dlpk\ArcGISObjectDetector.py", line 215, in vectorize
polygon_list, scores, classes = self.child_object_detector.vectorize(**pixelBlocks)
File "C:\Users\Administrator\AppData\Local\Programs\ArcGIS\Pro\bin\Python\envs\deeplearning282\Lib\site-packages\arcgis\learn\models\_inferencing\_yolov3_inference.py", line 334, in vectorize
for feature_idx in range(len(image_bbox.data[0])):
File "C:\Users\Administrator\AppData\Local\Programs\ArcGIS\Pro\bin\Python\envs\deeplearning282\Lib\site-packages\fastai\vision\image.py", line 378, in data
lbls = np.array([o.data for o in lbls]) if lbls is not None else None
File "C:\Users\Administrator\AppData\Local\Programs\ArcGIS\Pro\bin\Python\envs\deeplearning282\Lib\site-packages\torch\tensor.py", line 486, in __array__
return self.numpy()
TypeError: can't convert CUDA tensor to numpy. Use Tensor.cpu() to copy the tensor to host memory first.
Python Raster Function's .vectorize() method returned nothing.
Failed to execute (DetectObjectsUsingDeepLearning).
Failed at 13 August 2021 17:32:01 (Elapsed Time: 23,14 seconds)

 

Image below is the information from nvidia-smi:

Capture4.jpg

 

Hopefully anybody here can help me. Thank you. 

0 Kudos
1 Solution

Accepted Solutions
bhariadi
New Contributor III

The problem solved.

Only need to update the ArcGIS Pro version to 2.9.

Cuda enabledCuda enabled

 

 

 

View solution in original post

0 Kudos
4 Replies
bhariadi
New Contributor III

Update information (but it is not solving my previous problem yet):

I tried to train my data trough Jupyter Notebook & the CUDA worked. But when I used Detect Object Using Deep Learning tool in ArcGIS Pro, the error still persistent.

See the picture below.

Capture6.jpg

0 Kudos
bhariadi
New Contributor III

The problem solved.

Only need to update the ArcGIS Pro version to 2.9.

Cuda enabledCuda enabled

 

 

 

0 Kudos
shlirkd
New Contributor II

Hello bhariadi,

I am not able to get the yolov3 model working using the detect objects using deep learning method.

How did you do that? 

Regards

D

0 Kudos
bhariadi
New Contributor III

Hi,

Usually. I am using Jupyter Notebook not via ArcGIS Pro.

Regards,

 

0 Kudos