https://community.esri.com/t5/arcgis-image-analyst-questions/deep-learning-torch-cuda-outofmemoryerror-cuda-out/m-p/1580498#M799 <P>Hi&nbsp;<a href="https://community.esri.com/t5/user/viewprofilepage/user-id/814733">@PPRTDUA</a>&nbsp;,</P><P>Generally, out-of-memory issues occur when the batch size exceeds your GPU's capacity. Could you try running the tool with a lower batch size and check if any other processes are utilizing the GPU on your machine?</P> Thu, 30 Jan 2025 17:17:43 GMT ShivaniPathak 2025-01-30T17:17:43Z https://community.esri.com/t5/arcgis-image-analyst-questions/deep-learning-torch-cuda-outofmemoryerror-cuda-out/m-p/1572655#M784 <P>The problem occured when i used the&nbsp;multiresolution-object-detection-with-text-sam, it says error like this, how can i fix it?</P><P>&nbsp;</P><P>Starting execution<BR />Clearing CUDA cache...<BR />CUDA cache cleared.<BR />'Detect Objects Using Deep Learning' will run 5 times.<BR />Processing cell size: 0.1<BR />Processing cell size: 0.15<BR />Processing cell size: 0.2<BR />Repairing geometry...<BR />Geometry repaired.<BR />Deleting rows with areas &lt; 15 and areas &gt; 4500<BR />Rows deleted.<BR />Running Pairwise Buffer...<BR />Running Pairwise Dissolve...<BR />Pairwise Dissolve completed.<BR />Running Spatial Join...<BR />Spatial Join completed.<BR />Running Pairwise Dissolve - Mean Confidence...<BR />Pairwise Dissolve - Mean Confidence completed.<BR />Running feature classification by area...<BR />Running Regularizing Rectangular Footprints with tolerance 0.5...<BR />Regularizing Rectangular Footprints with tolerance 0.5...<BR />Regularizing Rectangular Footprints with tolerance 0.5 completed.<BR />Running Regularizing Rectangular Footprints with tolerance 1...<BR />Regularizing Rectangular Footprints with tolerance 1...<BR />Regularizing Rectangular Footprints with tolerance 1 completed.<BR />Running Regularizing Rectangular Footprints with tolerance 1.5...<BR />Regularizing Rectangular Footprints with tolerance 1.5...<BR />Regularizing Rectangular Footprints with tolerance 1.5 completed.<BR />Running Regularizing Rectangular Footprints with tolerance 2.5...<BR />Regularizing Rectangular Footprints with tolerance 2.5...<BR />Regularizing Rectangular Footprints with tolerance 2.5 completed.<BR />Running Regularizing Rectangular Footprints with tolerance 3.5...<BR />Regularizing Rectangular Footprints with tolerance 3.5...<BR />Regularizing Rectangular Footprints with tolerance 3.5 completed.<BR />Running Regularizing Rectangular Footprints with tolerance 5...<BR />Regularizing Rectangular Footprints with tolerance 5...<BR />Regularizing Rectangular Footprints with tolerance 5 completed.<BR />Deleting temporary outputs...<BR />Temporary outputs deleted.<BR />Running Merge...<BR />Running Merge for ['C:\\Users\\USER\\Documents\\ArcGIS\\Projects\\Kerja Kerja Kerja\\Kerja Kerja Kerja.gdb\\sds_20_50', 'C:\\Users\\USER\\Documents\\ArcGIS\\Projects\\Kerja Kerja Kerja\\Kerja Kerja Kerja.gdb\\sds_20_100', 'C:\\Users\\USER\\Documents\\ArcGIS\\Projects\\Kerja Kerja Kerja\\Kerja Kerja Kerja.gdb\\sds_20_150', 'C:\\Users\\USER\\Documents\\ArcGIS\\Projects\\Kerja Kerja Kerja\\Kerja Kerja Kerja.gdb\\sds_20_250', 'C:\\Users\\USER\\Documents\\ArcGIS\\Projects\\Kerja Kerja Kerja\\Kerja Kerja Kerja.gdb\\sds_20_350', 'C:\\Users\\USER\\Documents\\ArcGIS\\Projects\\Kerja Kerja Kerja\\Kerja Kerja Kerja.gdb\\sds_20_500']...<BR />Merge completed.<BR />Processing cell size: 0.25<BR />Processing extent: 1435233,7007148 9426028,69129479 1435778,57441159 9426483,61758766 PROJCS["DGN_1995_UTM_Zone_49S",GEOGCS["GCS_DGN_1995",DATUM["D_Datum_Geodesi_Nasional_1995",SPHEROID["WGS_1984",6378137.0,298.257223563]],PRIMEM["Greenwich",0.0],UNIT["Degree",0.0174532925199433]],PROJECTION["Transverse_Mercator"],PARAMETER["False_Easting",500000.0],PARAMETER["False_Northing",10000000.0],PARAMETER["Central_Meridian",111.0],PARAMETER["Scale_Factor",0.9996],PARAMETER["Latitude_Of_Origin",0.0],UNIT["Meter",1.0]]<BR />Detecting objects using deep learning...<BR />Traceback (most recent call last):<BR />File "&lt;string&gt;", line 371, in execute<BR />File "C:\Program Files\ArcGIS\Pro\Resources\ArcPy\arcpy\ia\Functions.py", line 4197, in DetectObjectsUsingDeepLearning<BR />return Wrapper(<BR />^^^^^^^^<BR />File "C:\Program Files\ArcGIS\Pro\Resources\ArcPy\arcpy\sa\Utils.py", line 45, in swapper<BR />result = wrapper(*args, **kwargs)<BR />^^^^^^^^^^^^^^^^^^^^^^^^<BR />File "C:\Program Files\ArcGIS\Pro\Resources\ArcPy\arcpy\ia\Functions.py", line 4182, in Wrapper<BR />result = arcpy.gp.DetectObjectsUsingDeepLearning_ia(<BR />^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^<BR />File "C:\Program Files\ArcGIS\Pro\Resources\ArcPy\arcpy\geoprocessing\_base.py", line 532, in &lt;lambda&gt;<BR />return lambda *args: val(*gp_fixargs(args, True))<BR />^^^^^^^^^^^^^^^^^^^^^^^^^^^^<BR />arcgisscripting.ExecuteError: A raster error has occurred. The messages that follow will provide more detail.<BR />Python raster function is unable to vectorize the data. [Failed to generate table]<BR />Python raster function is unable to vectorize the data.<BR />Traceback (most recent call last):<BR />File "C:\Users\USER\AppData\Local\Temp\ArcGISProTemp9596\TextSAM.dlpk\TextSAM.py", line 476, in vectorize<BR />masks, _, _ = self.mask_generator.predict_torch(<BR />^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^<BR />File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\torch\utils\_contextlib.py", line 115, in decorate_context<BR />return func(*args, **kwargs)<BR />^^^^^^^^^^^^^^^^^^^^^<BR />File "C:\Users\USER\AppData\Local\Temp\ArcGISProTemp9596\TextSAM.dlpk\segment-anything\segment_anything\predictor.py", line 229, in predict_torch<BR />low_res_masks, iou_predictions = self.model.mask_decoder(<BR />^^^^^^^^^^^^^^^^^^^^^^^^<BR />File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\torch\nn\modules\module.py", line 1501, in _call_impl<BR />return forward_call(*args, **kwargs)<BR />^^^^^^^^^^^^^^^^^^^^^^^^^^^^^<BR />File "C:\Users\USER\AppData\Local\Temp\ArcGISProTemp9596\TextSAM.dlpk\segment-anything\segment_anything\modeling\mask_decoder.py", line 94, in forward<BR />masks, iou_pred = self.predict_masks(<BR />^^^^^^^^^^^^^^^^^^^<BR />File "C:\Users\USER\AppData\Local\Temp\ArcGISProTemp9596\TextSAM.dlpk\segment-anything\segment_anything\modeling\mask_decoder.py", line 127, in predict_masks<BR />src=src + dense_prompt_embeddings<BR />~~~~^~~~~~~~~~~~~~~~~~~~~~~~~<BR />torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 10.60 GiB (GPU 0; 6.00 GiB total capacity; 11.87 GiB already allocated; 0 bytes free; 11.94 GiB reserved in total by PyTorch) If reserved memory is &gt;&gt; allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF<BR />Python Raster Function's .vectorize() method returned nothing.<BR />Failed to execute (DetectObjectsUsingDeepLearning).</P><P><BR />Failed to execute (MultiScaleDL).</P> Sun, 05 Jan 2025 10:36:30 GMT https://community.esri.com/t5/arcgis-image-analyst-questions/deep-learning-torch-cuda-outofmemoryerror-cuda-out/m-p/1572655#M784 PPRTDUA 2025-01-05T10:36:30Z https://community.esri.com/t5/arcgis-image-analyst-questions/deep-learning-torch-cuda-outofmemoryerror-cuda-out/m-p/1580498#M799 <P>Hi&nbsp;<a href="https://community.esri.com/t5/user/viewprofilepage/user-id/814733">@PPRTDUA</a>&nbsp;,</P><P>Generally, out-of-memory issues occur when the batch size exceeds your GPU's capacity. Could you try running the tool with a lower batch size and check if any other processes are utilizing the GPU on your machine?</P> Thu, 30 Jan 2025 17:17:43 GMT https://community.esri.com/t5/arcgis-image-analyst-questions/deep-learning-torch-cuda-outofmemoryerror-cuda-out/m-p/1580498#M799 ShivaniPathak 2025-01-30T17:17:43Z https://community.esri.com/t5/arcgis-image-analyst-questions/deep-learning-torch-cuda-outofmemoryerror-cuda-out/m-p/1583952#M809 <P>Hi&nbsp;<a href="https://community.esri.com/t5/user/viewprofilepage/user-id/814733">@PPRTDUA</a>&nbsp;,<BR /><BR />There could be multiple ways to avoid that error.</P><P>&nbsp;</P><P>1- As&nbsp;<a href="https://community.esri.com/t5/user/viewprofilepage/user-id/485093">@ShivaniPathak</a>&nbsp; suggested please try to decrease your batch size</P><P>2- I would try to run the tool with less resolutions or a smaller extent. Currently I see that you are processing 4 different resolutions and that could be utilizing lots of gpu memory.</P><P>Also what version of ArcGIS Pro are you currently running?</P><P>&nbsp;</P> Tue, 11 Feb 2025 12:15:24 GMT https://community.esri.com/t5/arcgis-image-analyst-questions/deep-learning-torch-cuda-outofmemoryerror-cuda-out/m-p/1583952#M809 Rami-Alouta 2025-02-11T12:15:24Z