3 ms·
Interesting project. It really should be compared against some databases of what crops are actually being grown to get a better sense of what one might take awa
by ChuckMcM 8y ago
Interesting project. It really should be compared against some databases of what crops are actually being grown to get a better sense of what one might take away from it. Just spot checking local cases around the Bay area seem to be about 40 - 50% accurate. But there are also issues where crops are rotated so I can imagine one issue would be getting crop images from different seasons could throw it off as well.
That it presents numbers as if it were 100% accurate, is an issue for me.
- onesoilteam 8y agoThank you! We compared the results of our algorithm to the datasets we were able to get and the accuracy of crop classification, or F1 score, is 0.91. This is stated in the "How it works" section of the map: https://map.onesoil.ai/2018?about https://map.onesoil.ai/2018?about. Hope this answer is helpful :)