I developed a software many years ago to do exactly this. GIS software is no longer an option when you get to the 1000s of plots. A field like show here would take just a couple of minutes with the custom software. For more complicated grids you could split it (and rejoin) at selected rows or columns. A key to a good fit is the transform/deformation type: I found bilinear is superior to projective.
We do this on thousands of plots regularly, but if the spacing is tight, it can be cumbersome to try to separate exactly, so I usually try to build a little extra spacing into the trials when possible, or just extract from the centers of canopies. If you have links to other helpful tools, let me know!
Hi Travis! Thanks for the great tutorial for QGIS, I am a beginner of this, so I have no idea how do you get the threshold layer, 0 value layer, 1 value layer, can you tell me more about it? Thanks again!!
Hi, this is part of a larger playlist: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-obRng3LLkRs.html Specifically, this video might be useful: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-uklnnLmfN6U.html
Hi, I usually don't want to cut my raster images up into large numbers of smaller images, usually it is more manageable to just extract data from each cell in the grid but keep a single raster. But here's something to try: 1. Open both the raster and vector files in QGIS. 2. Make sure that the vector file is in the same coordinate reference system (CRS) as the raster file. You can check this by right-clicking on the layer and selecting "Properties" > "General" > "Coordinate Reference System". 3. If the vector file is not in the correct CRS, you can change it by right-clicking on the layer and selecting "Set Layer CRS". 4. Select the raster layer and go to "Raster" > "Extraction" > "Clip Raster by Mask Layer". 5. In the "Clip Raster by Mask Layer" dialog, select the vector layer as the "Mask Layer". 6. If necessary, you can adjust the other settings, such as the output file format and the output extent. 7. Click "Run" to clip the raster file based on the vector grid. 8. The clipped raster layer will be added to the QGIS project, and you can view it in the Layers panel. 9. You can also save the clipped raster as a new file by right-clicking on the layer and selecting "Export" > "Save As". In the "Save Raster Layer As" dialog, choose the output file format and location, and click "Save".
Hi Travis! Thanks for sharing your knowledge in those items. I am also experimenting in those matters. After getting the Threshold to separate plant from soil, I have converted the raster file to a vector file using the Polygonize tool (under Raster Menu) to isolate each plant and make a calculation of how many plants I have in the plot. Did you tryed with this tool too or do you use other command/tool/function to summarize the amount of plant on a plot? Thanks a lot in advance.
Hi Eduardo, nice find! Yes this is something I have done for stand counts of plants in a plot. Unfortunately for us, our planting equipment doesn't drop seeds at equal spacing so we get clumps growing together that aren't separated. But in fields with uniform crop spacing and few weeds, this would work well.
This video series are excellent, it was really help me to get proper idea about my final year research. Thank you very much Can i contact you for get more instructions about drone mapping and NDVI time series for paddy varieties
These videos have been a massive help with my work! Do you know if there's a simple way to assign a row and column number to the attribute table for each plot? All of our plots use a grid reference starting at y1:x1 in the bottom left corner. Solving this issue will make sorting the data lot easier.
Hi, I think the easiest way to do that would be after exporting to Excel, R, etc. The cell IDs are numbered in a logical way, so once they are in Excel you can add columns for x and y and set up a repeating pattern (1, 2, 3, 4... etc., then make it repeat using an equation [such as D1=E1] and drag). There are a lot of little autofill and equation tricks that can make it so that you can do this even for a huge field in 30 seconds or so. In R, you can do the same kinds of things using things like the c() and rep() functions. Sorry if this is a little over-abbreviated, it is a little hard to write out!
@@travisparkerplantscience Hi, thanks for the response. This is currently the way I’m sorting it. We have a CSV file for all our trials that has column and row numbering so as long as it’s sorted by row and then by column and the numbers starts in the bottom left corner, it should, in theory, be in the right order. The issue is that I’m moving plots around and this messes up the sequence. Do you have an address I can email you on? I might be able to explain my problem better where I can show you some images?
@@s7ewie Hi, yeah the ordering should be correct before moving plots around, but for sure check it (especially if you have some outlier plots, most fields do). As far as customization and moving the polygons around, I would just recommend being really careful if you aren't doing it systematically. But otherwise you should just be able to similarly rearrange/customize your outputs or your input csv. If you are going to use the same field for many flights or whatnot, you can put a new column in your csv that is the id of the attribute table and sort by that for future data insertion
@@travisparkerplantscience The main issue is that I don’t want to include any tractor wheeling in the plot boundary’s as they skew the data. So I need to apply a horizontal and a vertical buffer. I can do this with the “buffer” tool but I need a greater buffer for the top and bottom than I do for the sides so I’ve been either creating a single row and “copy and moving” it or creating a grid with smaller dimensions and shifting each row manually and this tends to mess up the sequence. It’s not an ideal solution but I can’t currently find a better way. There is a “Plot Boundry” plugin but I can’t get it to work.
@@s7ewie I'm glad you used that buffer function, that is perfect. It's interesting that you say that shifting rows in the grid would mess up the sequence, it makes sense for copy + pasting a single row but moving the polygons shouldn't change their IDs. Are you sort of leap-frogging rows or just spacing them out a little extra? You could also try eliminating the tractor wheel areas by masking, such as shown in this video (which you have likely seen anyway): ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-uklnnLmfN6U.html
Hi travis, nice tutorial!! just a question, my ebee sq comes with just 4 multispectral bands, I dont have the blue one, is there any way to get it by separating RGB image?
Hi, if you are using a Sequoia, I would usually NDVI (RED-NIR)/(RED+NIR), which is done automatically in Pix4d. You would need to click in the processing options, in the index section, "merge tiles" and make sure "NDVI" is checked, which it should be by default. More information here: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-n3af0I_QYxA.html If needed, you could pull the RGB bands from the RGB camera on the Sequoia. I would either use those images for everything or nothing, instead of using the sq multispectral bands for R and G and the rolling shutter RGB for Blue only.
@@capsulaproducciones3370 Yes, since you have a camera capable of capturing the bands needed for NDVI, I would recommend that! I think of ExG as being a similar index for folks who only have RGB cameras
Hello Travis , we are trying to count agave plants , but we are usign a RGB camera dron ,(no multiepectral) , for doing the orthomosic ,so all I am getting is a tiff orthomosic where all bands are on one file orthomosic ? the tiif file gives me 4 bands but I cant really clasify them on E,G,B as you are doing for procesing , can you give me any advice ?
Yes, although I haven't loaded my pointclouds into QGIS for data extraction. Check out the following for more info: www.pointsnorthgis.ca/blog/viewing-point-clouds-qgis/#:~:text=Starting%20version%203.18%2C%20released%20in,cloud%20in%202D%20or%203D.
