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: Pooling layer error when importing YOLOV5 in TIDL

Part Number: TDA4VM

Hello team,

I tried importing YOLOv5 model in TIDL-RT. (SDK :08_02_00_05)

I observed the following error.

kindly let me know the solution.

Below is the import file and prototxt file used.

IMPORT FILE


modelType = 2
numParamBits = 8
numFeatureBits = 8
#quantizationStyle = 3
#quantizationStyle = 2
inputNetFile = "../../test/testvecs/models/public/onnx/yolo5_11.onnx"
outputNetFile = "../../test/testvecs/config/tidl_models/onnx/tidl_net_yolo5.bin"
outputParamsFile = "../../test/testvecs/config/tidl_models/onnx/tidl_io_yolo5_"
inDataNorm = 1
#inMean = 0 0 0
#inScale = 0.003921568627 0.003921568627 0.003921568627
inDataFormat = 1
inWidth = 640
inHeight = 640
inNumChannels = 3
numFrames = 1
inData = "../../test/testvecs/config/detection_list_yolo5.txt"
perfSimConfig = ../../test/testvecs/config/import/device_config.cfg
inElementType = 0
outDataNamesList = "/model.17/cv3/act/Mul_output_0,/model.20/cv3/act/Mul_output_0,/model.23/cv3/act/Mul_output_0"
metaArchType = 6
metaLayersNamesList = "../../test/testvecs/config/import/public/onnx/tidl_import_yolo5_metaarch.prototxt"
postProcType = 2

PROTOTXT FILE

name: "yolo_v5"
tidl_yolo {
name: "yolo_v5"
in_width: 640
in_height: 640
#10,13, 16,30, 33,23, 30,61, 62,45, 59,119, 116,90, 156,198, 373,326
yolo_param {
input: "/model.17/cv3/act/Mul_output_0"
anchor_width: 19
anchor_width: 44
anchor_width: 38
anchor_height: 27
anchor_height: 40
anchor_height: 94
}
yolo_param {
input: "/model.20/cv3/act/Mul_output_0"
anchor_width: 96
anchor_width: 86
anchor_width: 180
anchor_height: 68
anchor_height: 152
anchor_height: 137
}
yolo_param {
input: "/model.23/cv3/act/Mul_output_0"
anchor_width: 140
anchor_width: 303
anchor_width: 238
anchor_height: 301
anchor_height: 264
anchor_height: 542
}
detection_output_param {
num_classes: 9
share_location: true
background_label_id: 0
nms_param {
nms_threshold: 0.45
top_k: 5
}
code_type: CENTER_SIZE_EXP
keep_top_k: 5
confidence_threshold: 0.25
}
}

Regards,
Padmasree N.