Himax examples for Seeed Grove Vision AI Module V2

This is a repository which step by step teaches you how to build your own examples and run on Seeed Grove Vision AI Module V2. Finally, teach you how to restore to the original factory settings and run SenseCraft AI from Seeed Studio.

Outline

• How to build the firmware?
  • Build the firmware at Linux environment
  • Build the firmware at MacOS environment
  • Build the firmware at Windows environment
• How to flash the firmware?
  • System Requirement
  • Flash Image Update at Linux Environment by python code
  • Flash Image Update at Windows Environment by python code
  • Flash Image Update at Linux Environment
  • Flash Image Update at Windows Environment
  • Flash using Edge Impulse CLI tools
• How to restore to the original factory settings?
  • Linux Environment
  • Windows Environment

• How to add support for raspberry pi camera?

• How to use CMSIS-NN at the project?

• How to use CMSIS-DSP at the project?

• How to use CMSIS-CV at the project?

  • please clone the project by following command to download CMSIS-CV library

    git clone --recursive https://github.com/HimaxWiseEyePlus/Seeed_Grove_Vision_AI_Module_V2.git
    

• How to run Edge Impulse Example: standalone inferencing using Grove Vision AI Module V2 (Himax WiseEye2)?

  • ei_standalone_inferencing

  • ei_standalone_inferencing_camera

• FAQ

How to build the firmware?

This part explains how you can build the firmware for Grove Vision AI Module V2.

Build the firmware at Linux environment

Note: The following has been tested to work on Ubuntu 20.04 PC

• Step 1: Install the following prerequisites

  sudo apt install make
  

• Step 2: Download Arm GNU Toolchain (arm-gnu-toolchain-13.2.rel1-x86_64-arm-none-eabi.tar.xz)

  cd ~
  wget https://developer.arm.com/-/media/Files/downloads/gnu/13.2.rel1/binrel/arm-gnu-toolchain-13.2.rel1-x86_64-arm-none-eabi.tar.xz
  

• Step 3: Extract the file

  tar -xvf arm-gnu-toolchain-13.2.rel1-x86_64-arm-none-eabi.tar.xz
  

• Step 4: Add arm-gnu-toolchain-13.2.Rel1-x86_64-arm-none-eabi/bin/: to PATH
  (hint: the path about [$HOME/arm-gnu-toolchain-13.2.Rel1-x86_64-arm-none-eabi/bin/] should set your own path, you can not just copy and paste)

  #this is just the example, you can not just copy and paste !!
  export PATH="$HOME/arm-gnu-toolchain-13.2.Rel1-x86_64-arm-none-eabi/bin/:$PATH"
  

• Step 5: Clone the following repository and go into Seeed_Grove_Vision_AI_Module_V2 folder

  git clone --recursive https://github.com/HimaxWiseEyePlus/Seeed_Grove_Vision_AI_Module_V2.git
  cd Seeed_Grove_Vision_AI_Module_V2
  

• Step 6: Compile the firmware

  cd EPII_CM55M_APP_S
  make clean
  make
  

• Output elf file: ./obj_epii_evb_icv30_bdv10/gnu_epii_evb_WLCSP65/EPII_CM55M_gnu_epii_evb_WLCSP65_s.elf
• Step 7: Generate firmware image file

  cd ../we2_image_gen_local/
  cp ../EPII_CM55M_APP_S/obj_epii_evb_icv30_bdv10/gnu_epii_evb_WLCSP65/EPII_CM55M_gnu_epii_evb_WLCSP65_s.elf input_case1_secboot/
  ./we2_local_image_gen project_case1_blp_wlcsp.json
  

• Output firmware image: ./output_case1_sec_wlcsp/output.img

Back to Outline

Build the firmware at MacOS environment

Note: The steps are almost the same as the Linux environment except Step 1 and Step 7.

• Step 1:
  • You should make sure your make is using GNU version make not BSD version make.

    make --version
    

    
  • If it is not GNU make , you should download it by following command.

    brew install make
    

  • After installation, you can access it with the command gmake to avoid conflicts with the default make.

    gmake
    

  • So, you should create an alias in your shell configuration file (like .bash_profile or .zshrc):

    alias make='gmake'
    

• Step 7: Generate firmware image file (using ./we2_local_image_gen_macOS_arm64 for MacOS)

  cd ../we2_image_gen_local/
  cp ../EPII_CM55M_APP_S/obj_epii_evb_icv30_bdv10/gnu_epii_evb_WLCSP65/EPII_CM55M_gnu_epii_evb_WLCSP65_s.elf input_case1_secboot/
  ./we2_local_image_gen_macOS_arm64 project_case1_blp_wlcsp.json
  

Back to Outline

Build the firmware at Windows environment

• Step 1: Install the make command for prerequisites , you can reference here
• Step 2: Download Arm GNU Toolchain arm-gnu-toolchain-13.2.rel1-mingw-w64-i686-arm-none-eabi.zip
• Step 3: Extract the file

  tar -xvf arm-gnu-toolchain-13.2.rel1-mingw-w64-i686-arm-none-eabi.zip
  

• Step 4: Add arm-gnu-toolchain-13.2.rel1-mingw-w64-i686-arm-none-eabi/bin/ to your environment variables
  • option1: add the gnu-toolchain to your environment variables manually
  • option2: add the gnu-toolchain by command (hint: the path about [location of your gnu-toolchain-13.2 ROOT] should set your own path, you can not just copy and paste)

    #this is just the example, you can not just copy and paste !!
    setx PATH "%PATH%;[location of your gnu-toolchain-13.2 ROOT]\arm-gnu-toolchain-13.2.rel1-mingw-w64-i686-arm-none-eabi\bin"
    

• Step 5: Clone the following repository and go into Seeed_Grove_Vision_AI_Module_V2 folder

  git clone --recursive https://github.com/HimaxWiseEyePlus/Seeed_Grove_Vision_AI_Module_V2.git
  cd Seeed_Grove_Vision_AI_Module_V2
  

• Step 6: Compile the firmware

  cd EPII_CM55M_APP_S
  make clean
  make
  

• Output elf file: ./obj_epii_evb_icv30_bdv10/gnu_epii_evb_WLCSP65/EPII_CM55M_gnu_epii_evb_WLCSP65_s.elf
• Step 7: Generate firmware image file

  cd ../we2_image_gen_local/
  cp ../EPII_CM55M_APP_S/obj_epii_evb_icv30_bdv10/gnu_epii_evb_WLCSP65/EPII_CM55M_gnu_epii_evb_WLCSP65_s.elf input_case1_secboot/
  we2_local_image_gen project_case1_blp_wlcsp.json
  

• Output firmware image: ./output_case1_sec_wlcsp/output.img

Back to Outline

How to flash the firmware?

This part explains how you can flash the firmware to Grove Vision AI Module V2.

System Requirement

• Driver
  • If you find that the Grove Vision AI V2 is not recognised after connecting it to your computer, you should install the driver which can reference here.

1. Grove Vision AI Module V2
2. Connection cable
   • Micro usb cable: connect to EVB (as Power/UART)
3. Software Tools Serial terminal emulation application
   • In the following description, TeraTerm and Minicom will be used.
     • Serial terminal emulation application Setting
       • Baud Rate 921600 bps
       • Data 8 bit
       • Parity none
       • Stop 1 bit
       • Flow control none
       • please check xmodem protocol is supported.
     • Minicom (for Linux PC)
       • Install minicom command

         sudo apt-get install minicom
         

       • Burn application to flash by using xmodem send application binary.
         • Minicom will extra install "lrzsz" package to support xmodem protocol
         • If you did not have “lrzsz” instruction, please install by following instruction.

           sudo apt-get install lrzsz #(to support xmodem protocol)
           

       • Open the permissions to access the device

         sudo setfacl -m u:[USERNAME]:rw /dev/ttyUSB0
         # in my case
         # sudo setfacl -m u:kris:rw /dev/ttyACM0
         

         
       • Open minicom

         sudo minicom -s
         

         
       • Setup serial port and COM Port name
         • Tips for finding the COM Port name.
           • You can use google chrome to connect to Seeed SenseCraft AI, select Grove Vision(V2) and press Connect.
           • Then, you would get the COM Port name.
     • TeraTerm (for Windows PC)
       • Setup serial port

Back to Outline

Flash Image Update at Linux Environment by python code

• Prerequisites for xmodem
  • Please install the package at xmodem/requirements.txt

    pip install -r xmodem/requirements.txt
    

• Disconnect Minicom
• Make sure your Seeed Grove Vision AI Module V2 is connect to PC.
• Open the permissions to acceess the deivce

sudo setfacl -m u:[USERNAME]:rw /dev/ttyUSB0
# in my case
# sudo setfacl -m u:kris:rw /dev/ttyACM0

• Open Terminal and key-in following command
• port: the COM number of your Seeed Grove Vision AI Module V2, for example,/dev/ttyACM0
• baudrate: 921600
• file: your firmware image [maximum size is 1MB]

  python3 xmodem/xmodem_send.py --port=[your COM number] --baudrate=921600 --protocol=xmodem --file=we2_image_gen_local/output_case1_sec_wlcsp/output.img
  
  # example:
  # python3 xmodem/xmodem_send.py --port=/dev/ttyACM0 --baudrate=921600 --protocol=xmodem --file=we2_image_gen_local/output_case1_sec_wlcsp/output.img
  

• model(optional): you can burn multiple models "[model tflite] [position of model on flash] [offset]"
  • Position of model on flash is defined at the code ~/tflm_yolov8_od/common_config.h (take tflm_yolov8_od for example)

  python3 xmodem/xmodem_send.py --port=[your COM number] --baudrate=921600 --protocol=xmodem --file=we2_image_gen_local/output_case1_sec_wlcsp/output.img --model="model_zoo/tflm_yolov8_od/yolov8n_od_192_delete_transpose_0xB7B000.tflite 0xB7B000 0x00000"
  
  # example:
  # python3 xmodem/xmodem_send.py --port=/dev/ttyACM0 --baudrate=921600 --protocol=xmodem --file=we2_image_gen_local/output_case1_sec_wlcsp/output.img --model="model_zoo/tflm_yolov8_od/yolov8n_od_192_delete_transpose_0xB7B000.tflite 0xB7B000 0x00000"
  

• It will start to burn firmware image.
• Please press reset buttun on Seeed Grove Vision AI Module V2.
• It will success to run the algorithm.

Back to Outline

Flash Image Update at Windows Environment by python code

• Prerequisites for xmodem
  • Please install the package at xmodem/requirements.txt

    pip install -r xmodem/requirements.txt
    

• Disconnect Tera Term
• Make sure your Seeed Grove Vision AI Module V2 is connect to PC.
• Open CMD and key-in following command
• port: the COM number of your Seeed Grove Vision AI Module V2
• baudrate: 921600
• file: your firmware image [maximum size is 1MB]

  python xmodem\xmodem_send.py --port=[your COM number] --baudrate=921600 --protocol=xmodem --file=we2_image_gen_local\output_case1_sec_wlcsp\output.img 
  # example:
  # python xmodem\xmodem_send.py --port=COM123 --baudrate=921600 --protocol=xmodem --file=we2_image_gen_local\output_case1_sec_wlcsp\output.img 
  

• model: you can burn multiple models "[model tflite] [position of model on flash] [offset]"
  • Position of model on flash is defined at the code ~/tflm_yolov8_od/common_config.h (take tflm_yolov8_od for example)

  python xmodem\xmodem_send.py --port=[your COM number] --baudrate=921600 --protocol=xmodem --file=we2_image_gen_local\output_case1_sec_wlcsp\output.img --model="model_zoo\tflm_yolov8_od\yolov8n_od_192_delete_transpose_0xB7B000.tflite 0xB7B000 0x00000"
  
  # example:
  # python xmodem\xmodem_send.py --port=COM123 --baudrate=921600 --protocol=xmodem --file=we2_image_gen_local\output_case1_sec_wlcsp\output.img --model="model_zoo\tflm_yolov8_od\yolov8n_od_192_delete_transpose_0xB7B000.tflite 0xB7B000 0x00000"
  

• It will start to burn firmware image automatically.
• Please press reset buttun on Seeed Grove Vision AI Module V2.
• It will success to run the algorithm.

Back to Outline

Flash Image Update at Linux Environment

Following steps update application in the flash.

• Step 1: Open Minicom, setup serial port and COM Port name-> connect to Grove Vision AI Module V2. (Please reference the minicom part of System Requirement)
• Step 2: Hold down any key on the keyboard (except the Enter key) and press the reset button to reset Grove Vision AI Module V2 and the startup options will be displayed.
• Step 3: Press button “1” and Grove Vision AI Module V2 will enter receiving mode after then.
  
• Step 4: Press Ctrl+A on keyboard to enter minicom menu, and then press s on keyboard to upload file and select xmodem.
• Step 5: Select the firmware image at Seeed_Grove_Vision_AI_Module_V2\we2_image_gen_local\output_case1_sec_wlcsp\output.img and press enter to burn.
• Step 6: After burning is compelete, press any key to be continue.
• Step 7: Then, you will see the message "Do you want to end file transmission and reboot system? (y)" is displayed. Press button y to restart.
• Step 8: You will see the uart on minicom which is runing your algorithm.

Back to Outline

Flash Image Update at Windows Environment

Following steps update application in the flash.

• Step 1: Open TeraTerm and select File -> New connection, connect to Grove Vision AI Module V2.
• Step 2: Hold down any key on the keyboard (except the Enter key) and press the reset button to reset Grove Vision AI Module V2 and the startup options will be displayed.
• Step 3: Press button “1” and Grove Vision AI Module V2 will enter receiving mode after then. Select target flash image(output.img) by File->Transfer->XMODEM->Send.
• Step 4: After the firmware image burning is completed, the message "Do you want to end file transmission and reboot system? (y)" is displayed. Press button y to restart.
• Step 5: You will see the uart on TeraTerm which is runing your algorithm.

Back to Outline

Flash using Edge Impulse CLI tools

This method works on all supported operating systems (Windows/Linux/MacOS...)

• Step 1: Install the Edge Impulse CLI tools
• Step 2: Open any system terminal and run the following command

  himax-flash-tool -d WiseEye2 -f <path_to_four_firmware_img_file>
  

• Step 3: Wait until you see the following message:

  [HMX] Press **RESET** to start the application...
  [HMX] Firmware update completed
  

Note: if the flashing process hangs, just cancel it (Ctrl+C) and start once again.

Back to Outline

How to restore to the original factory settings

Linux Environment

• Update the flash image Seeed_SenseCraft_AI*.img to Grove Vision AI Module V2 and press reset buttun.

• Disconnect the Minicom:

  • Please press Ctrl+A on keyboard and press z on keyboard to go to the menu of minicom.
  • Then, press q on keyboard to quit with no reset minicom, and press yes to leave.

• Open the permissions to acceess the deivce

  sudo setfacl -m u:[USERNAME]:rw /dev/ttyUSB0
  # in my case
  # sudo setfacl -m u:kris:rw /dev/ttyACM0
  

  

• After doing the above steps, you can run the SenseCraft AI on Grove Vision AI Module V2.

  1. Introduction : https://wiki.seeedstudio.com/grove_vision_ai_v2/
  2. Connect Grove Vision AI Module to NB USB port
  3. Open "Google Chrome" browser
  4. Open SenseCraft Homepage
  5. Select "Grove Vision AI(WE2)" and connect (serial port)

Back to Outline

Windows Environment

• Update the flash image Seeed_SenseCraft AI*.img to Grove Vision AI Module V2 and press reset buttun.
• Disconnect the TeraTerm.
• After doing the above steps, you can run the SenseCraft AI on Grove Vision AI Module V2.
  1. Introduction : https://wiki.seeedstudio.com/grove_vision_ai_v2/
  2. Install CH343 UART driver (CH343SER.ZIP) (Optional)
  3. Connect Grove Vision AI Module to NB USB port
  4. Open "Microsoft Edge" browser
  5. Open SenseCraft Homepage
  6. Select "Grove Vision AI(WE2)" and connect (serial port)

Back to Outline

How to add support for raspberry pi camera?

You can reference the scenario app allon_sensor_tflm , allon_sensor_tflm_freertos and tflm_fd_fm. Take allon_sensor_tflm for example, you should only modify the allon_sensor_tflm.mk from cis_ov5647 to cis_imx219 or cis_imx477.

#CIS_SUPPORT_INAPP_MODEL = cis_ov5647
CIS_SUPPORT_INAPP_MODEL = cis_imx219
#CIS_SUPPORT_INAPP_MODEL = cis_imx477

So that, it can support cis_imx219 or cis_imx477 camera.

Back to Outline

How to use CMSIS-NN at the project?

• Modify the setting at the makefile
  • Enable the flag LIB_CMSIS_NN_ENALBE to build CMSIS-NN library

    LIB_CMSIS_NN_ENALBE = 1
    

  • You can reference the scenario app example about allon_sensor_tflm_cmsis_nn which is the example running the model without passing vela and using the CMSIS-NN library.
    • Change the APP_TYPE to allon_sensor_tflm_cmsis_nn at the makefile

      APP_TYPE = allon_sensor_tflm_cmsis_nn
      

Back to Outline