AI 模型转换脚本

gistfile1.txt

#mtk caffe 模型转换
python scripts/caffe2pb_inference.py \
--model=models/resnet/ResNet-50-model.caffemodel \
--proto=models/resnet/ResNet-50-deploy.prototxt

python scripts/caffe2pb_inference.py \
--model=models/mobilenet/mobilenet_v2.caffemodel \
--proto=models/mobilenet/mobilenet_v2_deploy.prototxt

#华为 caffe 模型转换
./cngen_linux --prototxt mobilenet_v2_deploy.prototxt --model mobilenet_v2.caffemodel

#华为 tensorflow 模型转换
cd cpu2ipu && sh run.sh
./pb_to_offline --graph=../model/facenet_ipu.pb --param_file=../model/facenet_model.txt

#mtk tensorflow to tflite
#inception_v3
toco --graph_def_file=/home/tflite/neuropilot_converter/tf2tflite/sample_models/inception_v3.pb \
		--input_format=TENSORFLOW_GRAPHDEF \
		--output_format=tflite \
		--output_file=INCEPTION_V3.TFLITE \
		--inference_type=FLOAT \
		--inference_input_type=FLOAT \
		--input_arrays=input \
		--output_arrays=InceptionV3/Predictions/Reshape_1 \
		--input_shapes=1,299,299,3
#
toco --graph_def_file=/home/tflite/neuropilot_converter/tf2tflite/sample_models/resnet34.pb \
		--input_format=TENSORFLOW_GRAPHDEF \
		--output_format=tflite \
		--output_file=resnet34.TFLITE \
		--inference_type=FLOAT \
		--inference_input_type=FLOAT \
		--input_arrays=Placeholder \
		--output_arrays=loss \
		--input_shapes=1,224,224,3
#拿到模型input output
bazel  build tensorflow/tools/graph_transforms:summarize_graph

./bazel-bin/tensorflow/tools/graph_transforms/summarize_graph --in_graph=/home/kenwood/snpe-1.17.0/inception_v3.pb

bazel run tensorflow/tools/benchmark:benchmark_model -- --graph=/home/kenwood/snpe-1.17.0/inception_v3.pb --show_flops --input_layer=input --input_layer_type=float --input_layer_shape=-1,299,299,3 --output_layer=InceptionV3/Predictions/Reshape_1

bazel run tensorflow/tools/benchmark:benchmark_model -- --graph=/home/kenwood/snpe-1.17.0/resnet34.pb --show_flops --input_layer=Placeholder --input_layer_type=float --input_layer_shape=1,224,224,3 --output_layer=loss