Converting MobileNetv2.tflite model into tengine format

VIM3
#1

Hi,
As I was not successful in converting MobileNet SSD model, so I have temporarily moved on to a classification model.
I trained a model and converted it to tflite format. During my conversion of my model into .tmfile, I am getting the following error.

Tengine-Error-1
Tengine-Error-11440×189 25.1 KB

#2

@Akkisony Can you share you tflite model ?

#3

@Frank Google Drive: Sign-in

Please find the model in the above link. Hope you can download the model.
Thanks for your help! :slight_smile:

#4

@Frank Did you check the tflite model? Were you able to generate the .tmfile? If yes, please let me know where am I going wrong. Thank you in advance!

#5

@Akkisony I have consulted the technical staff of tengine and need time to confirm

#6

@Frank Thank you for your reply. Hope to get an answer soon! :slight_smile:

#7

@Akkisony Tengine not support tflite

#8

@Frank But, I saw it in the official website that there is support for tflite.
Link: Model Convert Tool — Tengine documentation

Tengine-Error-6
Tengine-Error-61087×582 52.9 KB

#9

@Akkisony Tengine’s technicians told me that I need to convert to onnx to continue the conversion

#10

I’m trying to convert MobilenetV2 frozen graph into *tmfile format and getting error:

---- Tengine Convert Tool ---- 

Version     : v1.0, 16:55:24 Aug 25 2022
Status      : float32
These ops are not in tensorflow serializer: 
[ NoOp ]
[ Enter ]
[ Switch ]
[ AddV2 ]
[ NextIteration ]
[ Merge ]
[ TensorListFromTensor ]
[ TensorListReserve ]
[ Pack ]
[ TensorListSetItem ]
[ LogicalAnd ]
[ LoopCond ]
[ TensorListGetItem ]
[ Unpack ]
[ Fill ]
[ Range ]
[ Tile ]
[ Slice ]
[ Exit ]
[ StopGradient ]
[ TensorListStack ]
[ FusedBatchNormV3 ]
[ NonMaxSuppressionV5 ]
[ GatherV2 ]
[ Select ]
[ Shape ]
[ Assert ]
[ Where ]
[ Size ]

Create graph failed
#11

Upd. tried to convert tensorflow efficiendet model, getting the same error