This document mainly introduces how to use the SDK to convert the user's own model and apply it to the NPU.

1. Train your own yolov3 model. The training method and process can refer to the official: Darknet Yolo Page, here we use the officially trained weights based on the coco data set.

2. Prepare SDK, app warehouse, and demo warehouse.

Please refer to the SDK, app and demo documents respectively for how to obtain the corresponding code:

1. NPU SDK Usage
2. NPU Prebuilt Demo Usage
3. Application Source Code

The conversion is performed under the SDK.

$ cd {workspace}/SDK/acuity-toolkit/conversion_scripts

1. Modify NAME:

NAME=mobilenet_tf --> NAME=yolov3

2. Comment out Tensorflow:

$convert_tf \
    --tf-pb ./model/mobilenet_v1.pb \
    --inputs input \
    --input-size-list '224,224,3' \
    --outputs MobilenetV1/Logits/SpatialSqueeze \
    --net-output ${NAME}.json \
    --data-output ${NAME}.data

Modify to,

#$convert_tf \
#    --tf-pb ./model/mobilenet_v1.pb \
#    --inputs input \
#    --input-size-list '224,224,3' \
#    --outputs MobilenetV1/Logits/SpatialSqueeze \
#    --net-output ${NAME}.json \
#    --data-output ${NAME}.data

3. Comment Darknet:

#$convert_darknet \
#    --net-input xxx.cfg \
#   --weight-input xxx.weights \
#    --net-output ${NAME}.json \
#    --data-output ${NAME}.data

Modify to,

$convert_darknet \
    --net-input path/to/yolov3.cfg \
   --weight-input path/to/yolov3.weights \
    --net-output ${NAME}.json \
    --data-output ${NAME}.data

1. Modify NAME:

NAME=mobilenet_tf --> NAME=yolov3

2. Modify regression parameters:

     --channel-mean-value '128 128 128 128' \

Modify to,

     --channel-mean-value '0 0 0 256' \

3. Modify validation_tf.txt:

Replace the image inside,

$ cat ./data/validation_tf.txt
./space_shuttle_224.jpg, 813

Modify to,

path/to/416x416.jpg

The picture resolution here is the same as the configuration in the yolo cfg file.

4. Modify quant type:

    --quantized-dtype asymmetric_affine-u8 \

Modify to,

    --quantized-dtype dynamic_fixed_point-i8 \

1. Modify NAME:

NAME=mobilenet_tf --> NAME=yolov3

2. Modify regression parameters:

     --channel-mean-value '128 128 128 128' \

Modify to,

     --channel-mean-value '0 0 0 256' \

3. Modify the RGB channel order:

default channel is RGB,

    --reorder-channel '0 1 2' \

Modified to BGR,

    --reorder-channel '2 1 0' \

4. Specified board model

VIM3:

    --optimize VIPNANOQI_PID0X88  \

VIM3L:

    --optimize VIPNANOQI_PID0X99  \

1. Compile:

$ bash 0_import_model.sh && bash 1_quantize_model.sh  && bash 2_export_case_code.sh

2. case code:

The converted code is in the nbg_unify_yolov3 directory,

$ ls {workspace}/SDK/acuity-toolkit/conversion_scripts/nbg_unify_yolov3
BUILD  main.c  makefile.linux  nbg_meta.json  vnn_global.h  vnn_post_process.c  vnn_post_process.h  vnn_pre_process.c  vnn_pre_process.h  vnn_yolov3.c  vnn_yolov3.h  yolov3.nb  yolov3.vcxproj

This part of the code is carried out in the aml_npu_app repository. Enter the directory of detect_yolo_v3.

$ cd {workspace}/aml_npu_app/detect_library/model_code/detect_yolo_v3
$ ls
build_vx.sh  include  Makefile  makefile.linux  nn_data  vnn_yolov3.c  yolo_v3.c  yolov3_process.c

1. Replace vnn_yolov3.h , vnn_post_process.h, vnn_pre_process.h generated by SDK.

$ cp {workspace}/SDK/acuity-toolkit/conversion_scripts/nbg_unify_yolov3/vnn_yolov3.h {workspace}/aml_npu_app/detect_library/model_code/detect_yolo_v3/include/vnn_yolov3.h
$ cp {workspace}/SDK/acuity-toolkit/conversion_scripts/nbg_unify_yolov3/vnn_post_process.h {workspace}/aml_npu_app/detect_library/model_code/detect_yolo_v3/include/vnn_post_process.h
$ cp {workspace}/SDK/acuity-toolkit/conversion_scripts/nbg_unify_yolov3/vnn_pre_process.h {workspace}/aml_npu_app/detect_library/model_code/detect_yolo_v3/include/vnn_pre_process.h

2. Replace vnn_yolov3.c generated by SDK.

$ cp {workspace}/SDK/acuity-toolkit/conversion_scripts/nbg_unify_yolov3/vnn_yolov3.c {workspace}/aml_npu_app/detect_library/model_code/detect_yolo_v3/vnn_yolov3.c

1. Modify the class array:

static char *coco_names[] = {"person","bicycle","car","motorbike","aeroplane","bus","train","truck","boat","traffic light","fire hydrant","stop sign","parking meter","bench","bird","cat","dog","horse","sheep","cow","elephant","bear","zebra","giraffe","backpack","umbrella","handbag","tie","suitcase","frisbee","skis","snowboard","sports ball","kite","baseball bat","baseball glove","skateboard","surfboard","tennis racket","bottle","wine glass","cup","fork","knife","spoon","bowl","banana","apple","sandwich","orange","broccoli","carrot","hot dog","pizza","donut","cake","chair","sofa","pottedplant","bed","diningtable","toilet","tvmonitor","laptop","mouse","remote","keyboard","cell phone","microwave","oven","toaster","sink","refrigerator","book","clock"
,"vase","scissors","teddy bear","hair drier","toothbrush"};

According to your training data set settings, if it is a coco data set, there is no need to modify it.

2. Modify yolo_v3_post_process_onescale:

Modify num_class:

int num_class = 80;

The num_class here is the same as the number of classes in the training set.

3. Modified post-processing function yolov3_postprocess.

Modify num_class and size[3].

int num_class = 80;
int size[3]={nn_width/32, nn_height/32,85*3};

• The num_class here is the same as the number of classes in the training set.
• The size[2] here is equal to (num_class + 5) * 3.

Use the build_vx.sh script to compile the yolov3 library,

$ cd {workspace}/aml_npu_app/detect_library/model_code/detect_yolo_v3
$ ./build_vx.sh

The generated library is in the bin_r directory

$ ls {workspace}/aml_npu_app/detect_library/model_code/detect_yolo_v3/bin_r
libnn_yolo_v3.so  vnn_yolov3.o  yolo_v3.o  yolov3_process.o

1. Replace yolov3 library

$ cp {workspace}/aml_npu_app/detect_library/model_code/detect_yolo_v3/bin_r/libnn_yolo_v3.so {workspace}/aml_npu_demo_binaries/detect_demo_picture/lib/libnn_yolo_v3.so

2. Replace nb file

VIM3:

$ cp {workspace}/SDK/acuity-toolkit/conversion_scripts/nbg_unify_yolov3/yolov3.nb {workspace}/aml_npu_demo_binaries/detect_demo_picture/nn_data/yolov3_88.nb

VIM3L:

$ cp {workspace}/SDK/acuity-toolkit/conversion_scripts/nbg_unify_yolov3/yolov3.nb {workspace}/aml_npu_demo_binaries/detect_demo_picture/nn_data/yolov3_99.nb

How to run the replaced aml_npu_demo_binaries on the board, please refer to:NPU Prebuilt Demo Usage.