Our new customers often ask us how much data they need to collect and how often they need to collect it.
This blog will outline everything you need to know about collecting data with GreenView.
What we mean by data
When we talk about GreenView ‘data’ we’re talking about the footage of your farm that you capture with a GoPro camera (often when you’re out performing jobs like mulching, mowing or spraying). The GreenView system takes this data and crunches the numbers into interactive dashboards and maps to give you useful insights on your crops.
When to collect data
The frequency of data collection is driven by the questions and problems your specific farm is facing.
For example, if you are a berry grower who is required to supply weekly data for yield forecasting, then weekly data collection makes sense.
If you are a grape grower wishing to understand how much fruit is in your vineyard prior to thinning, then this is a single point in time and you could just do one data collection. If you then wanted to assess how much fruit is left after thinning, you would perform a second data collection.
How much data to collect
The question of how much data to collect can be approached in two ways - either ad hoc as you’re out and about, or tailored to a certain problem you’re experiencing.
Collecting ad hoc data
The beauty of the GoPro is that it can be attached to farm equipment that is already making passes over the farm performing other operations. For instance if you need to mow grass, and will be doing every row, then you can attach the GoPro and easily capture data for every row while you’re at it. This gives the GreenView system plenty of information to provide broad insights on crop health and variability.
Collecting tailored data
The second approach to data collection is for specific capture when trying to gather information on a particular problem. In these instances a greatly reduced sample size can be collected, only focusing on the key area. This approach is also useful if your business is located in an area of slow internet speeds, to reduce the size of the videos that need to be uploaded.
For this tailored approach, firstly consider the size of the management unit you need to make a decision on. A rough rule of thumb is that for many block level decisions, a sample size of around 10% of the block would be adequate. So if your block has 100 rows in it, then capturing 10 rows in that block would be a robust sample size for decision making.
If however, there is the opportunity for differential management practices within a block, then more footage would need to be collected so that you have sufficient data. As a rough rule of thumb you would want to capture at least 10% of each of the different management units that the decision will be based on to capture crop variability.
Actionable insights from GreenView
Within 48 hours of uploading your data to GreenView you’ll be able to download interactive dashboards and maps full of actionable insights specific to your farm and crops.
This includes information on:
WHERE THE FUTURE MEETS THE FARM
The Bitwise Blog focuses on issues impacting Machine Learning in agriculture. We bring you industry news and insights for your business.