/
AI model convert tool: Caffe

AI model convert tool: Caffe

Owned by Rie Sakito (崎戸 梨恵) (Unlicensed)
Last updated: Feb 20, 2025 by Tachibana Yuta (立花 裕太)
3 min read

 

Preparation

Please refer here for the environment construction procedure.

Please note that the following environment settings must be performed every time the CV Tool container is started.

  • For ambaCV2X camera

$ cd /home/cvtool $ source setup_env.sh

  • For ambaCV5X camera

$ cd /home/cvtool $ source setup_env.sh ipro-ambaCV5X

 

Convert sample model

Change directory.

$ cd /home/cvtool/conversion/caffe

Download sample model.

$ wget https://github.com/shicai/MobileNet-Caffe/raw/master/mobilenet.caffemodel $ mv mobilenet.caffemodel sample/mobilenet_v1/models

Conver the model.

$ ./caffe_conversion.sh setting.conf


The model after conversion is output to the following directory.

  • For ambaCV2X camera

${OUTPUT_DIR}/${NET_NAME}/${PARSER_OPTION}/[model name]

  • For ambaCV5X camera

${OUTPUT_DIR}/${NET_NAME}_ambaCV5X/${PARSER_OPTION}/[model name]

Conversion

  1. Copy /home/cvtool/conversion/caffe .

$ cd /home/cvtool/conversion $ cp -r caffe foo $ cd foo

  1. Change the parameter of "setting.conf" according to the model to be converted.

  2. Convert the model.

$ ./caffe_conversion.sh setting.conf

 

setting.conf

From v1.20, parameter "CAVALRY_VER" is removed that existed in setting.conf until v1.19.
If you use setting.conf from v1.19 or earlier, please remove "CAVALRY_VER" from it.

# Network Name NET_NAME=mobilenetv1 # Path to Deploy Prototxt DEPLOY_PROTOTXT=./sample/mobilenet_v1/mobilenet_deploy.prototxt # Path to Directory for (Deploy) Caffe Models MODEL_DIR=./sample/mobilenet_v1/models # Path to Directory for DRA Images DRA_IMAGE_DIR=../dra_img # Path to Directory for Output Data OUTPUT_DIR=./out # Quantization Mode # FIX8 : Fixed-point 8bit # FIX16 : Fixed-point 16bit # MIX : FIX8/FIX16 mixed PARSER_OPTION=FIX8 # Input Data Format (0:NHWC, 1:NCHW) IN_DATA_FORMAT=1 # Input Data Channel IN_DATA_CHANNEL=3 # Input Data Width IN_DATA_WIDTH=224 # Input Data Height IN_DATA_HEIGHT=224 # Input Data Mean Vector or Name of .binaryproto IN_MEAN=103.94,116.78,123.68 # Input Data Scale # IN_SCALE=1/Scale IN_SCALE=58.823529411 # RGB or BGR (0:RGB, 1:BGR) IS_BGR=1 # Input Layer Name IN_LAYER=data # Output Layers Name OUT_LAYER=mbox_loc,mbox_conf_flatten # Unique preprocess # if use im2bin -> NONE # if use unique preprocess -> script path PREPRO=NONE PREPRO_ARG="" # Input file data format IN_DATA_FILEFORMAT=0,0,0,0 # Transpose indices(NONE:without transpose , 0,3,1,2:transpose (EX)) IN_DATA_TRANSPOSE=NONE

  • NET_NAME: The name of network.

    • Any name can be set.

  • DEPLOY_PROTOTXT: Path to deploy prototxt file

  • MODEL_DIR: Path to directory which includes caffemodel

    • All caffemodels under the converted directory

  • DRA_IMAGE_DIR: Path to directory which includes image files for optimizing quantization

    • Please put the directory image files for training. Recommended number of image files is 100 to 200.

    • Image file format should be the format that are supported by OpenCV (Ex. JPEG, PNG and etc…).

    • Any resolution is supported.

  • OUTPUT_DIR: Path to directory to which the converted data is placed

  • PARSER_OPTION: Quantization mode

    • Select from FIX8/FIX16/MIX (FIX8/FIX16 mixed).

  • IN_DATA_CHANNEL: Number of input image channel for target model

  • N_DATA_WIDTH: Width of input image for target model

  • IN_DATA_HEIGHT: Height of input image for target model

  • IN_MEAN: Normalization parameter (mean) of input image

    • It can be set by numerical value or .binaryproto file

    • Please refrain from using space between “,” as shown below if using numerical value.
      IN_MEAN=127.5,127.5,127.5

    • Please set a path to the file as shown below when setting by .binaryproto file.
      IN_MEAN=./model/mean.binaryproto

  • IN_SCALE: Normalization parameter (scale) of input image

    • Please refrain from using space between “,” and only use “,” to separate values when setting different values for each channel.

  • IS_BGR: Format of input image (RGB or BGR)

  • IN_LAYER: The name of input layer for target network

    • In the converted model, the name of input layer changes to “${IN_LAYER}_0” .
      Therefore, “_0” needs to be added to the name of input layer, when the converted model executes AdamApp.

    • When the following symbols are contained in the name of input layer, conversion may not be successful.
      : | ; , ‘

  • OUT_LAYER: The name of output layer for target network

    • If two or more layers exists, separate layers by “,”.

    • When the following symbols are contained in the name of output layer, conversion may not be successful.
      : | ; , ‘

  • PREPRO: Path of preprocessing script (python script)

    • Refer to “/home/cvtool/ common/prepro.py” for how to create a script.

  • PREPRO_ARG: Argument of preprocessing script (python script)

  • IN_DATA_FILEFORMAT: Input data format

    • examples: uint8->0,0,0,0, float32->1,2,0,7, float16->1,1,0,4

    • When the value of IN_DATA_FILEFORMAT changes from “0,0,0,0”, setting PREPRO is needed.

  • IN_DATA_TRANSPOSE: Specify when performing TRANSPOSE on the input data


If input layer is not defined in deployed prototxt file, add the layer as shown below.

 

Related content