Hello I'm Marc Degroseilliers from Technical Support here at Pix4D. Today we're talking about Sequoia. Sequoia is the multispectral camera from Parrot and it is typically used for agriculture
projects and most agriculture project the aim is to measure the reflectance of the different regions of the field and this is done by using a specific camera and typically there are two types of
cameras that are going to be used either modified cameras or multispectral cameras. What's the difference between those? The modified camera is just an regular RGB camera where the filters in
front of the lens are going to be a swapped out so they're going to install new filters which is going to change the wavelength that are measured by the camera so instead of RGB typically it's
going to be near infrared read and green for example and two things to keep in mind about these modified cameras is that they are broadband what I mean by this is that for agriculture project
typically you're interested that the reflectance at a very specific wavelengths see 719 nanometers and these this camera so then you want the the reflectance at this wavelength and the modified
camera is going to measure the energy in an interval around this wavelength which is going to be let's say larger than the other camera in the multispectral camera so that's what we mean by
broadband is going to integrate over a wider interval the amount of energy in this interval. The other thing that modified cameras do is that they're going to apply digital filters what this
means is that and the sensor is going to record some data and then before it is saved in the memory of the camera the SD card for example some digital filters are going to be applied this means
that the data is going to be changed these filters are applied by the camera firmware so that these images that were meant to be taken RGB and they look better they're more visually pleasing so
for example the green really looks like the green that was taken during when the shot was taken and but these digital filters they modify the data and so they make it more difficult to interpret
it afterwards This is not the case for the second type of camera which are multispectral cameras because typically each sensor is going to measure just one wavelength and these digital filters
are not going to be applied because these cameras were specifically designed to record the data as it was measured and the sensor they're typically going to be also a narrowband so because we're
interested that the reflectance at this 719 nanometers the sensor is going to measure the information just in a narrow interval around this wavelength. Another thing about multispectral cameras
is that since they were designed and engineered for agriculture purposes they will typically write all the necessary radiometric data in the exif of the images so these can be taken into account
and you can compute the reflectance from the images that were taken. So this is where I introduce Sequoia. It's in two parts there's the camera and there's the sunshine sensor the camera as four
narrowband sensors and one RGB sensor at the bottom here and there's also the sunshine sensor which is going to measure the illumination and in the sunshine sensor there is also the gps module
and the IMU. This is what sequoia looks like. About the sunshine sensor it's going to record the the amount of irradiance coming from the Sun for each picture and for each band and what this will
allow to do is to normalize the data between the different shots what I mean by this is that the amount of elimination changes during the flight so that all of a sudden at the end of light there
is more light coming from the Sun and so and there's going to be more light more photons that are captured by the camera but then that doesn't mean that the reflectance of this region of the
field was higher just means that the total amount of photons that was emitted was higher so then this illumination sensor allows you to take this into account and to normalize the value of the
images amongst themselves. Once you receive sequoia in the mail probably the first thing you'll want to do is to mount it on your drone there are integration kits that are available for the Ebee
from Sensefly and also for the DJI phantom 3 and the 3dr solo. If you want to mount sequoia on your own drone which is not one of those drones or you don't want to use this integration kit you
should keep in mind that you will need to power sequoia, it requires five volts and 2.4 amps and so that is if you have an external battery and one thing that we want to bring to your attention
is that right now there are plans that are available on the internet for 3d printing a universal mount for Sequoia. When you put the camera and the Sunshine sensor on your drone it's important to
have the camera facing down so that you can measure your field and to have the sunshine sensor pointing in the opposite direction towards the sky. Once you've set this up you will want to connect
to sequoia to change the settings you can do this by connecting to sequoia via Wi-Fi and after the flight when you want to connect to sequoia to download the images you can either connect to it
the Wi-Fi or with the cable or you can also just take out the SD card if it the images were recorded on the SD card Let's see now on the computer what are the exact steps to take to connect to
sequoia via Wi-Fi and change the settings. Let me now connect the sequoia to the computer which will also power it (It should be recognized by the computer in a few seconds) There you go Let me
connect to sequoia via Wi-Fi and this is how I can change the settings of the Sequoia. So I open the browser, I go to the address that was written on the slides (here we are here). Here I can
choose how to record the images whether on the internal memory or on the SD card and I can also choose which sensors are going to be used to for example I just want a multispectral imagery or I
just need RGB and here in the capture window I can choose the mode so either a single picture time-lapse so a picture every X seconds GPS mode which will depend on the position of the Sequoia so
how far it has traveled or just auto mode with overlap here. So for example if I choose time-lapse mode then I have a handy calculator and I can write the speed of my drone the height and then
the overlap desired and it will compute so let's say 10 m/s and at 70 meters and I want an overlap of 80% (here you go). I click apply and now you see it it will take a picture every 1.3 seconds.
Now if I try, there is a lower limit so that let's say that I tried to put 0.3 then it will not accept this value because it's too low and it will bring it back to the minimum which in this case
is 1 so this minimum value will depend on two factors: it will depend on whether the pictures are safe to internal memory or to the SD card it's quicker to save to internal memory if you want to
take more pictures faster and also which sensors are activated. So for example if just the multispectral sensors are activated and you don't take our RGB imagery then you will be able to set a
lower time-lapse interval. Here that was for the first window now you can also go in the other tab so the current status of Sequoia you can look at the gallery so I've took a few pictures before
I changed the folder and you have the help and the settings where you can reset to factory settings for example. That's what it looks like when you connect in through the browser to change the
options and here you can also download the imagery but sequoia will also be recognized as a camera so if you go in the file in the folder manager then you can also look here and you see sequoia
which is going to be seen as a drive and you can see your imagery and download your imagery from here. Let's see how to create the project once we have the sequoia imagery so here I have my
images and what I like to do to sort them and put them in folders according to whether they're RGB or multispectral the easiest way to do this is just to sort by file type because here I
have my tiff imagery which is multispectral so I just put my folder and the jpeg images are the RGB (there you go) That will make my life a little bit easier now I can launch the mapper and
create a new project (Sequoia multispectral). So I will just create the multispectral project, add images as Sequoia multispectral all of them > next As you can see there is one camera that's
created for each of the multispectral sensors and they are connected by the rig geometry and click > next Looks good and the coordinate system is automatically detected > next and now I'm
going to choose AG multispectral because I'm processing the multispectral imagery if I was processing the RGB project and I would choose here Ag RGB Ag multispectral and finish. So here's my
project and now: about the calibration target if you are using version 2.2 of the software and firmware of the Sequoia is version 1.1 at least then the calibration images will be recognized
automatically and the region is selected and the calibration target is taken care of automatically. If this is not the case then you should do it manually. One important note to make sure that
it's done automatically is that the calibration target which is recognized is the Airinov one, if you're using another calibration target then it will not be recognized automatically and then you
will have to do it by hand as I can show you now: so I go in the processing options > I go in the DSM, Orthomosaic and Index for the third step and in the index calculator. I have here my four
sensors and I need to calibrate each of them I do this in the following way I choose calibrate I browse to find in my multispectral imagery now as you can see I'm calibrating the green sensor so
I take my green calibration image and I draw a region inside of the reflectance region of the target and this number here it's written on my target I know what it is this case it's 179 and I
click calibrate OK and I would do this for all of the all of the sensors I would need to do this three more times but I'm not going to do this here then I click ok and since I chose the AG
multispectral template everything is pre-selected, all the processing options and I can just click start That's how you start your project with the sequoia imagery. In many of these agriculture
projects you might want to use a calibration target to be able to compare datasets between that were flown during different days. To use a calibration target it's important to keep a few things
in mind: first thing is that the target should be flat on the ground not tilted and if it's a handheld target such as the Airinov one you will want to take pictures of it either right before or
right after the flight and we recommend taking a few pictures just to make sure that at least one set of imagery is good and can be used for calibration one common pitfall is that the the target
is going to be either underexposed or overexposed. If the target is overexposed it cannot be used for calibration so it's important to make sure that one set of imagery looks good and and
typically overexposure will happen if the background is very dark and so the camera tries to compensate for it so if you want to avoid this you can see what works well for your project but we
would recommend trying to use some kind of light background for the for the the calibration shot. It's also important to avoid shadows on the target as this would change the reflectance of the
target region and as a side note starting from version 2.2 of the mapper the Airinov target is recognized automatically for sequoia projects so you don't need to input the different values for
the reflectance of the different bands you just import the images in your project and the calibration shots are recognized automatically and your project is calibrated by itself. There's no need
to input the values manually. So then once you're done flying you will extract the data either from the internal memory of Sequoia or from the SD card, download it on your computer and
process the data in Pix4Dmapper. One thing to keep in mind is that you should have two projects: one for the RGB imagery and one for the multispectral imagery and this should be two projects and
you should just import the correct imagery in each project. When you're processing there are two templates to use so for the multispectral imagery it's the AG multispectral and for the RGB
imagery the template is called AG RGB and in this case It's important to make sure that in the camera model the linear rolling shutter model is selected. Once you've done the processing you will
get the output for your project so in the case of RGB imagery is going to be the orthomosaic and in the case of multispectral imagery it's going to be the reflectance maps from which you can get
the index maps and the prescription maps. To see how to go from reflectance to index maps to prescription map well this is what is done in the index calculator and we have another video that
explains this workflow so we encourage you to go have a look to see how to use the index calculator to get the data that you're looking for. That's all I wanted to say about Sequoia today, I hope
you found this video useful and I'll see you in the next video.
