请注意，本文内容源自机器翻译，可能存在语法或其它翻译错误，仅供参考。如需获取准确内容，请参阅链接中的英语原文或自行翻译。
器件型号：PROCESSOR-SDK-DRA8X-TDA4X
我有一个 onnx 模型、它的输入是单通道灰度图像、我将它导入到 TIDL 模型、并且我在 TDA4上推理、它的输出是错误的。
这是我的导入配置文件。
modelType = 2.
NumFrames = 1
numParamBits = 16
numFeatureBits = 16
量化样式= 3
inputNetFile ="/home/wanggang/workspace/superpoint/model/superonnx
outputNetFile ="/home/wanggang/workspace/superpoint/model/super.bin
outputParamsFile ="/home/wanggang/workspace/superpoint/model/super_
inFileFormat = 1.
inDataFormat = 0
InWidth = 640
InHeight = 480
InElementType = 1.
reszeWidth = 640
电阻高度= 480
inNumChannels = 1.
in均 值= 0
Inscale = 1
InData = filename.bin
PostProcType = 0
inDataNorm = 1.
foldedPreBnConv2D = 0
folddBnInConv2D = 1.
outDataNamesList ="128"
debugTraceLevel = 1.
writeTraceLevel = 3
