HUSKYLENSUploader

Windows

Install .Net Framework https://www.microsoft.com/en-US/download/details.aspx?id=56116
Run K-Flash.exe on administrator.
Select the firmware and serial port.
Click to flash.

Mac and Linux

We recommand to upload the firmware on windows using HUSKYLENSUploader as it has a GUI. If you want to upload it on Mac and Linux please following these instruction:

Install the USB Serial driver depending on your OS: https://www.silabs.com/products/development-tools/software/usb-to-uart-bridge-vcp-drivers
Install pip3 if you do not have in your OS

Install pip3 on MAC

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install.sh)"
brew install python3

install pip3 on Linux

sudo apt install python3-pip

Run the following code to install pyserial:

sudo pip3 install pyserial

Go to the HUSKYLENSUploader folder

cd HUSKYLENSUploader

Run the following code to update the firmware:

sudo python3 kflash.py -b 2000000 HUSKYLENSWithModelV0.4.7Stable.kfpkg

Disconnect and reconnect the USB to reboot the HUSKYLENS to make it a refresh start up.

Version Different

File name	Detail
HUSKYLENSWithModelVx.x.xStable.kfpkg	Normal firmware with models
HUSKYLENSVx.x.xStable.bin	Normal firmware without models
HUSKYLENSWithModelVx.x.xClass.kfpkg	Object classification firmware with models
HUSKYLENSVx.x.xClass.bin	Object classification firmware without models

What is different between with models and without?

Models stores the deep learning architecture and weights like MobileNet or YOLO. It is really large and cost a lot of time to upload.

Since the models not always change, firmware without models are provided to speed up the upload process.

Why object classification is separated?

Because object classification requires a lot of ram, which is not enough for other algorithm.