This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TDA4VM: Get output bin file in TDA4VM chip environment

Part Number: TDA4VM

Hello TI,

Is it possible to get the output bin file of inference in TDA4VM?

I have been trying to import BlitzNet (deep learning model) in TIDL, and I have tried to use the object detection post processing that is provided by TIDL. 

I tried all methods that TIDL provides (Object detection API, metaArch).

Found that all of the post processing methods doesn't work for this model. 

So I have cut the model right before post processing steps and imported in TIDL. (Trying to do post processing on CPU)

And it worked in emulation environment.

Config file for inference is as below.

inFileFormat    = 2

postProcType = 0

netBinFile      = "testvecs/config/tidl_models/tflite/tidl_net_blitznet.bin"

ioConfigFile   = "testvecs/config/tidl_models/tflite/tidl_io_blitznet1.bin"

inData  =   testvecs/config/detection_list.txt

outData =   testvecs/output/blitznet.bin

writeTraceLevel = 3

debugTraceLevel = 1

writeOutput = 2

However, in TDA4VM environment, it doesn’t work. Actully there was no explicit error message, but inference execution never ends (like infinite loop).

Is there have any way to get output tensor bin file in the TDA4VM inference process?