So... I want to be able to detect and ID seedlings of crops and weeds to advance the [open weed locator][https://github.com/geezacoleman/OpenWeedLocator] to make more cost effective the outphasing of synthetic systemic herbicides.
The project has plenty of room for advancement. One of those opportunities, I initialy thought, was to use a V3 NoIR camera ( I thought an obvious go-to for plant detection, as opposed to a standard camera), but I'm not so sure after checking on some of the info.
For starters, how are people calulating NDVI if the V3 NoIR camera, in my understanfing, returns the
| Red + NIR + IR | Green | Blue | bands, (no IR cut-off filter, right?)
as opposed to the normal one which is (IR filter doing it's thing, right?)
| Red | Green | Blue | bands
NDVI needs a separate Red and NIR bands.... which aren't separate on the V3· camera, right?
And the filters would only block some band for a specific sensor. i.e. the blue one blocks blue light, so the blue sensor value would be close to zero, but would leave the other's relatively unaffected, right?
Also, the NoIR V3 camera has no filters to be added or removed anyway, right?
So under the | Red+ NIR + IR | Green | Blue | bands premise, substituting | Red | OR | NIR |
by | Red+ NIR + IR | for indexes that don't depend on Red/NIR separation like TGI , CIg, GNDVI or VARI seems the most doable thing for getting any relevant indexes and plant / not-plant initial detection, right?
Does PlantCV support whatever the NoIR V3· RBPi camera produces anyway? (i guess it would be IR grayscale (discarding | Green | Blue | ) vs RGB ( | Red | Green | Blue | ) processing)
I can't seem to find somewhere this is explained (at least the sensor relative response vs wavelenght for each camera model, the least one could expect) and would greatly appriciate the help.
The project has plenty of room for advancement. One of those opportunities, I initialy thought, was to use a V3 NoIR camera ( I thought an obvious go-to for plant detection, as opposed to a standard camera), but I'm not so sure after checking on some of the info.
For starters, how are people calulating NDVI if the V3 NoIR camera, in my understanfing, returns the
| Red + NIR + IR | Green | Blue | bands, (no IR cut-off filter, right?)
as opposed to the normal one which is (IR filter doing it's thing, right?)
| Red | Green | Blue | bands
NDVI needs a separate Red and NIR bands.... which aren't separate on the V3· camera, right?
And the filters would only block some band for a specific sensor. i.e. the blue one blocks blue light, so the blue sensor value would be close to zero, but would leave the other's relatively unaffected, right?
Also, the NoIR V3 camera has no filters to be added or removed anyway, right?
So under the | Red+ NIR + IR | Green | Blue | bands premise, substituting | Red | OR | NIR |
by | Red+ NIR + IR | for indexes that don't depend on Red/NIR separation like TGI , CIg, GNDVI or VARI seems the most doable thing for getting any relevant indexes and plant / not-plant initial detection, right?
Does PlantCV support whatever the NoIR V3· RBPi camera produces anyway? (i guess it would be IR grayscale (discarding | Green | Blue | ) vs RGB ( | Red | Green | Blue | ) processing)
I can't seem to find somewhere this is explained (at least the sensor relative response vs wavelenght for each camera model, the least one could expect) and would greatly appriciate the help.
Statistics: Posted by dpicassom — Mon Jun 03, 2024 10:32 pm