Completely new to this...stand counts, plant heights, NDVI

I’m new to drones in general. I was looking @ the inspire…Curious about cameras etc. I do contract research for a large number of companies, therefore I have countless stand counts and plant height ratings in my test plots. Can the correct setup do this? If so, what setup? Most stand counts are conducted early season…VE-V6. Does it simply take the pics and then we count off the pics…or does a software count the plants within a specified area?Most plant heights are V6-V10. Also, what NDVI camera is the best?

1 Like

Basically, at the moment, yes. I don’t know what the Drone Deploy guys have planned, here, but I’m sure they’ll tell you - either in house algorithms or an API for 3rd party analytics like DataMapper, I’d imagine? Anyway, yes. Take pictures and Drone Deploy will stitch into orthomosaic. Then you need to count off this. If you have no other way then manually counting. We normally run the imagery through some CV ( computer vision) script in Mathcad which does boundary selection to isolate the plants and from there does a plant count. So, it can be automated. There are a few scripts for this kicking around on GITHUB if you are tech-savy. Or the DD guys might have something up their sleeve that they can do for you if you ask them nicely - they’re a good Dev team.

Well…yes. The standard I1 with x3 camera can do this fine. Flying a map at up to V6 at somewhere around 40m will give you the required results in imagery. There are also a couple of NDVI conversions available for the X3. AerialMediaPros is a popular choice, but it depends what you are looking to get for it. The Drone Deploy NDVI tool suite is very good, though, and works well with this camera.

The X5 camera only adds a little and IMHO for mapping I’d question if it justified the price. Particularly when adding 3rd party lenses. If I was spending that much on a platform I would not use an Inspire. Although you can always add extra batteries, camera etc to the Inspire if you so wished.

The biggest technical challenges are not with what is in the air but with what is on the ground. Patchy cloud causes big issues on NDVI analysis, so you need to be weather-aware. If you are going to run daily or weekly assessments then you will need ground control points (which the support guys can help you in processing). Then you need to make sure your plots are very well defined, well spaced, and if you intend to follow through with height analysis later in the growth period you will need to consider you boundary to avoid crop fall-over and encroachment on the edges. The CV algorithms look for crisp, clear, well-defined edges. (There is some smarts that can be added to this, but in a basic sense this is how it works). So neighbouring plants cannot touch. Weeds cause problems. Wind and falling over will greatly affect any height analysis, etc… The better you can set out and space your test plots the better your results will be. (It is also worth noting the valid method of extending the plots at the edges but discounting a boundary area to negate many of the effects mentioned above. This works, but is wasteful of you test plot site. You need to consider what will work best for you. )

5 Likes

Thanks! So is V6 about the cutoff for accurate plant heights? I’m new to NDVI also…it csn be used throughout the growing season to evaluate plant health from what I understand. …can it also be used late season for Stay Green ratings?

There are lots of variables in that, but basically: no. You can do it any time. However, the more established and / or thicker the crop / canopy cover the more issues you will have with using an SFM (Structure from Motion / Photogrammetry) technique. Laser / Radar solutions that can penetrate the canopy and give you a true ground level reading that you can then subtract from canopy height will always yeild superior results. There’s a load of papers on this. EG: 1 2

Yep. Again, several papers, but some of the best quick-intro stuff I’ve read is from Agribotix and Sensefly

The resounding theme you will see running through all of the papers and blog posts, though, and in line with our own in-house R&D findings where I work, is that the fundamental theories are sound but practical, real-world execution of the scientific method at a reasonable price-point is a lot harder to achieve. It’s why Rothamsted built this.

1 Like

ugh…makes sense…somewhat. Knowing what you know I need…what drone/cameras would you recommend?

Would the Matrice 100 be better than inspire?

Sorry to part Hijack this thread…when using third party NDVI lenses from say IRPro or Aerial Media Pros, does the narrower field of view cause problems building a grid in the app that will still produce enough overlap?

We have a guide for the modified cameras with some suggestions we’ve seen work. Others on the forum might have additional tips. http://support.dronedeploy.com/docs/ndvi-cameras

Thanks. Any suggestions on which mod fits in best with app would be good. Verging on ordering IRPro mod for Inspire, but worried that FOV is only 60 degrees.

Does anyone make a LiDAR for the Inspire or Matrice?

A little OT, but it’s your thread.

LiDAR on Inspire / Matrice?
Nope! Inspire is Prosumer and Matrice is for low-end R&D. LiDAR done right is 5-10X this price point or more. I don’t expect anyone will ever put it on one of those platforms although if you really wanted to burn some cash you could probably stick one of these (Phoenix Aerial) or these (Routescene) on a Matrice.

Would a Matrice be better than an Inspire?
Not really but it depends on the sensor payload you intend to use. Matrice gives you (a few) more options but they are not really built for fieldwork. If you want to stick to the DJI camp (and fair enough - good and easy to use flight controllers) then for optimal adaptability an S900 / S1000+ and Lightbridge is a very good platform. Your choice of camera, but you can then run Lidar (Someone not far from me runs one with a Lidarpod), multispec / hyperspec and standard stills / video cameras as well as a stack of other stuff. (Sure, you can do this on a Matrice but you’ll need to reinvent a few wheels, first). For your use case I’d be looking at a hexa or octo - slow, low, stable, minimal prop wash - and something like a Tetracam (There’s loads of options in this market, now, each with pros and cons) but it’s all going to come down to a balance between your budget and your required accuracy. For an S900 and Multispec you’ll be looking around $10k - double a fully-spec’d Inspire + NDVI camera. For the LiDAR’s you are looking at between double and ten times that. If you don’t need such high-quality, standardised and repeatable data then the Inspire might just do!

1 Like

Thanks Roger…been a great deal of help!

New here to Drone deploy. Not meaning to hujack or whatever. I have a direct question and you sound well educated on this.
If i have a camera 12mp (like on phantom 3 pro) that has aerial media pro NDVI filter VS a precision hawk camera Multispec (BGNIR) 14.2 mp how would the results in Drone deploy differ? Or in general any processing software?
Is there enough difference to justify buying the more expensive?
Drone deploy has not answered my emails on NDVI formulas ( if theirs is true or not, since they offer RGB NDVI???)

Hi,
Where did you send your emails? We try to answer all emails that arrive at support@droneseploy.com

i think i sent one to that email address and also info@dronedeploy.com… had one response and reply to the latter but nothing since.

should i email the support address again?

I can answer these questions here. I was just wondering why you didn’t receive a response.the modified camera won’t be as good as a real nir camera but it can get the job done for a lot of people. It will decrease the field of view and sometimes the modifications can decrease the sharpness of the photos. You need to fly with higher overlap and make sure your camera settings are right.
I think your results with either camera should be fine with dronedeploy. We don’t offer rgb ndvi, we offer ndvi coloring options that you can use if your imagery is NIR. If it isn’t NIR then the coloring doesn’t mean very much.

I might be biased but I think an expensive solution like precision hawk will be fully replaced by much more affordable options such as djis and others in the near future. It’s already sufficient for most people’s use cases.
You can use dronedeploy for free so you can always compare the two of you like.

1 Like

just bought a phantom pro and going to give drone deploy a try. I almost have my 333 exemption and will be commercial. DJI just does not have a fixed wing for large acreage. DJI is marketed to average consumers and can be used with different camera filters. Hawk is getting ready for the upcoming sense and avoid issues FAA will be addressing. I am afraid that will become an issue for legal operators with 333 exemptions. Afraid FAA will require all flights to have this sense and avoid. DJI would have a huge problem on their hands if this was the case. Ok, I also am a little biased. DJI may replace more expensive platforms but when Precision data counts it counts… A mining company that has millions of dollars in revenue would want a more accurate data acquisition i beleive.

I watched a youtube video today of drone deploy and i replied to it. the video said RGB images turned to NDVI… This confused me. As true NDVI is with NIR variations and the correct NIR-/+NIR algorithm. Does drone deploy use this? So I have done NDVI with gopros and i am looking to upgrade and try a camera that logs metadata such as phantom pro and use drone deploy. I am looking for precision resolution. Does drone deploy flight planner auto adjust for flying at lower alltitudes for better resolution? {lower flight speed and shutter speed)

Is there a phone number to call you guys? I dont want to annoy this forum or hijack it or whatever. first one i been on.
plus i have like a billion questions.

@chasemgray nailed it.

The simple answer is…

You get what you pay for!

The Multispec on the Lancaster is a calibrated and well integrated camera. MP comparison only gives you so much, although you do need to factor the extra time a lesser camera / platform means you will take on the job. Furthermore, Algo Market on Datamapper is going to give you lots of good addons for analysis of your data. But awesome though it is, the PH package isn’t cheap. All in, it’s around 7X, I think? (Disclosure: I :heart_eyes: PH stuff!)

Another way in which you are not comparing like-for-like is Lancaster (a fixed wing) vs DJI Quadcopter. If you want large acre coverage then I’d be looking at a fixed wing (but you don’t need to look far on here! ). But if you want something more akin to a vineyard / orchard or test plots then you need slow, steady, close up… multirotor (or ground bot).

The DroneDeploy NDVI toolkit also gives you the ability to leverage a lot of the functionality hard coded in PH’s offering, so it is more flexible…but less finely tuned.

If a quad would do then in all honesty I’d get a quad + NDVI + thermal + Multispec because then you can analyse the :hankey: out of stuff and still be less $$$ than a Lancaster.

If you want high-level precision, RTK positioning, calibrated multispec, etc then something like Sensefly or high-end Pixhawk platforms (which includes PH)

Re SaA… Yeah, PH have LATAS and their foot in the door with NASA but they are not the only kid on the block and NASA’s tech is vendor neutral. DJI also have a (baby) foot in that same door, but they have a much more international market and international focus. They are not ignoring SaA by any means, but their focus thus far has been building the consumer market for low altitude, low-speed multirotors. Here SaA is less of an issue. However, their focus is changing. A lot of guys in my neck of the woods also have SaA tech, but commercial adoption is still a couple of years away (perhaps a little longer allowing for legislators to catch up), by which time the tech and sensors would have moved on so much you’ll probably be looking for your next platform. So I wouldn’t worry about SaA yet.

More important right now is meeting site safety requirements. I know plenty of sites won’t let you near the place if you only have a non-redundant quadrotor and will seriously frown at you about the high energy ballistic implications of fixed wings. (One reason eBee wins so many orders)

So, what do you want to do with it?

Yes we get what we pay for and I see your point. We have been looking at a fixed wing and multirotor pair to use for work. I plan on agriculture, survey and map. Lancaster has been offered for an attractive price actually. They usually start around 30k and full package is 220k. I have an offer that is 16k ish. Honestly i like drone deploys tactics and where your going. I feel like your processing will boom here when it does get tuned.

Sensefly ebee is 22-30k and their multirotor is 44k.

Are phantoms considered redundant? i want to say no. WHat do you mean frown on high energy ballistic implications of fixed wings?

If you guys want input on some good defaults for the ag toolbox, so it is flexible but also has most of the same hard coded settings you get with PH, please provide suggestions and feedback. We’re building and prioritizing things based on customer feedback.

1 Like