What is the best way to count corn plants in a field? I have drone imagery, mostly infrared but some rgb.
This post is closed to further activity.
It may be an old question, an answered question, an implemented idea, or a notification-only post.
Please check post dates before relying on any information in a question or answer.
For follow-up or related questions, please post a new question or idea.
If there is a genuine update to be made, please contact us and request that the post is reopened.
It may be an old question, an answered question, an implemented idea, or a notification-only post.
Please check post dates before relying on any information in a question or answer.
For follow-up or related questions, please post a new question or idea.
If there is a genuine update to be made, please contact us and request that the post is reopened.
+60- Celebrity
- 3702 replies
It really depends on what your imagery looks like. If you can visually identify individual plants you could sample the RGB values of those pixels and then try to identify, and count, clumps of those through FME.
I'd try the RasterCellCoercer and then classify ranges of RGB (AttributeManager or AttributeRangeMapper), dissolve those pixels and that should do it.
I've done something similar for trees, out of a point cloud (so no RGB to work with) which worked ok except when tree canopies started to overlap.
+56- Safer
- 2554 replies
What stage of development are we talking here? If it's early germination then I think the @redgeographics approach would be good because it would be few spots of green in a mostly bare field. And with IR it would be even more distinct.
But if it's at the high growth point, then most of the field would be green and the density of plants such that I can't feel that separating out individual plants is practical.
At that stage I wonder if it would be easier to find bare earth instead, where growth is not taking place. Do you know the supposed plant spacing? I think say find the total area of the field and then the total area of bare earth using the detection methods described. Then assume say 3 plants per metre (one per foot), and calculate the potential plants from that given what you know about the cultivated field area.
I live in a rural area so I can try and find out what the locals here do.
If it is IR maybe take a look at the computer vision stuff in FME 2019 and train a model for object detection?