TDA4VM: TIDL object detection application usage with ONNX model (Proto) file : ./resnet18_opset9.onnx , TIDL network file : ./out/tidl_net.bin , and TIDL IO info file : ./out/tidl_io_buff

Part Number: TDA4VM

Tool/software:

Dear Team.

Based on the suggestion from the link https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1555548/tda4vm-how-to-generate-outputnetfile-and-outputparamsfile-for-onnx-tesorflow-using-tidl_model_import-out 

we have generated
ONNX model (Proto) file : ./resnet18_opset9.onnx
TIDL network file : ./out/tidl_net.bin
TIDL IO info file : ./out/tidl_io_buff

Now, we would like to use the above generated TIDL network file and io info files and run the object detection application.
cd /opt/vision_apps
source vision_apps_init.sh
./run_app_tidl_od.sh

Application fails with the below error.

Application logs:
Entered to app_verify_graph func ...
228.069719 s: VX_ZONE_ERROR: [tivxMemBufferAlloc:141] size is 0
228.069735 s: VX_ZONE_ERROR: [ownAllocReferenceBufferGeneric:349] Memory allocation failed
228.069744 s: VX_ZONE_ERROR: [ownGraphAllocateDataObject:1254] Memory allocation for replicated parameter parent object failed
228.069756 s: VX_ZONE_ERROR: [ OpenVxGraph ] Memory alloc for data objects failed
228.069764 s: VX_ZONE_ERROR: [ OpenVxGraph ] Graph verify failed
App Verify Graph Done!

Could you please help us to run the run_app_tidl_od.sh application with generated net and io files from the above link.
Note we have placed the tidl_net.bin and tidl_io_buff files in the respective directories as mentioned in the app_of.cfg file.

We are using:

J721E EVM board.

We are running on ubuntu 22.04

Any help or pointers to documentation/examples would be appreciated.

Thanks and Regards
Firdose