The ModelAccuracyTool-Armnn
is a program for measuring the Top 5 accuracy results of a model against an image dataset.
Prerequisites:
ArmnnConverter
can be used to convert a model to this format.ImageTensorGenerator
can be used to convert the test images to this format.Cmd: | ||
---|---|---|
-h | --help | Display help messages |
-m | --model-path | Path to armnn format model file |
-c | --compute | Which device to run layers on by default. Possible choices: CpuRef, CpuAcc, GpuAcc. Default: CpuAcc, CpuRef |
-d | --data-dir | Path to directory containing the ImageNet test data |
-i | --input-name | Identifier of the input tensors in the network separated by comma |
-o | --output-name | Identifier of the output tensors in the network separated by comma |
-v | --validation-labels-path | Path to ImageNet Validation Label file |
Example usage:
./ModelAccuracyTool -m /path/to/model/model.armnn -c CpuRef -d /path/to/test/directory/ -i input -o output -v /path/to/file/val.txt