Introduction to Yield Mapping
What is the most important time throughout the year on a farm? The harvest time, because it determines the profit that the farm can aspire to as a result of all efforts and investments made during the year. Today, combine harvesters are equipped with instruments that allow real-time visualization of the harvested yield, and an appreciation of its variability on a map. But there are some wineries that politically choose to do manual harvesting, and therefore by not using harvesting machines, they cannot get a yield map. Or there are farms that cannot afford to buy or use such equipment. And in these contexts where the combination of the use of computer vision, cameras, IoT and cloud solutions can provide the yield map.
Traditional Yield Mapping: Limitation and future
Yield mapping is only possible following the introduction of GPS. Thanks to GPS nowadays we can orient ourselves in space directly from our smartphones. But GPS is of fundamental importance for yield mapping, because it allows us to associate a given quantity collected in space with its relative coordinate. Below you will find the map created above a durum wheat field in Italy.
As you can see, a yield map is nothing more than individual GPS points that are, however, accompanied by the relative yield. This is raw data that needs to be further processed and cleaned up, in fact there are numerous problems in yield maps today, and they are:
Some data overestimate yield, for example, on durum wheat in the Marche there is typically a yield of 5 to 7.5 t/ha. Some yield maps have shown values well over 10 t/ha, and obviously these data need to be filtered
Some times due to lack of internet connection or that the RTK system is not working properly, some data are not georeferenced and therefore you have holes
Some companies have a manual collection policy and therefore these maps cannot be generated
So such data if taken as is cannot generate useful information for proper management. In fact, one needs specific tools that have been developed to be able to perform proficient processing.
But these instruments are very expensive or on certain farms cannot be used because they are too expensive. So should the farmer do without them? Actually there are systems, with a much lower cost than combines that can generate the yield map. We have developed one together with Iselqui Technologies s.r.l.
Computer Vision: the future yield mapping technology
We have developed a hardware-software solution that allow to map the yield of vineyard with a affordable price. We call it "FruitKount".
This new service is composed by:
Hardware, which can be mounted in every tractors, that acquired images of your vineyard at high resolution. Our AI Based Computer Vision Model can detect the number, size and weight of grapes. Moreover each image is georeferenced in order to create a variability yield map.
Software, all data acquired by our hardware is sent to our cloud infrastructure which the data are evaluated and archived. Moreover we have an web and mobile application that can be used to see the variability yield map.
Here below a yield map create by our system.
Right now the service is being tested at some farms and will potentially be made public in late September, which can be fully connected with our precision farming application.
Benefits of Computer Vision for Farmers
You may be wondering about the possible advantages of using this instrumentation in the viticultural context, below we make a list of the applications we have made so far:
Create prescription maps to differentiate fertilizer supply, water, pruning, tillage
Implementing separate harvest
Have reliable data in case of hailstorms or wildlife damage
Plan the personnel needed to accomplish the harvest
Identify areas of the field that exhibit low production, which may be an indicator of a fungal or virus attack
Here below some images where you can appreciate the accuracy of our AI Model.
White Grape
Red Grape
コメント