Artificial Intelligence & the Economy | Making an AI-based malicious weed detection application in under a week
Artificial Intelligence and the Economy features machine-learning computer models in Jamaica. These models are computer algorithms, or smart apps, that seek to give computers the ability to learn like children for a variety of tasks.
Here, we highlight how machine learning/artificial intelligence can be applied for small farmers. In this work, one can leverage the use of a machine-learning tool called KaggleAPI, complete with a type of template for a class of smart computer application called Convolutional artificial neural networks. This way one does not need to build this smart app from scratch, instead one may configure the smart Kaggle tool for the purpose of a task such as weed detection in crops.
No NEED to build apps from scratch
One can write basic artificial neural networks from scratch, but nowadays there are tools available that remove the need for the researcher/software person to write these models from scratch. We can now leverage powerful modern machine learning tools, that already come complete with templates or computer code structures that describe many types of artificial neural networks, such as Convolutional Artificial Neural Networks, which can be configured to be good at computer image tasks, such as weed classification/detection in crops.
Smart weed detection benefits
The Ministry of Agriculture lists agricultural growth problems, in relation to malicious plants/weeds. Although expensive drone services offer weed detection strategies, after studying Artificial Neural Networks for roughly two weeks, a junior software developer can devote as little as under a week of software development time, to develop a basic weed reporting and identification platform for small farmers. The first article in this column spoke about a curriculum for studying machine learning.
Anyway, new small farmers would be able to snap and upload images of plant species that appear to be causing issues, and quickly get back information about which weed type the plant is, and therefore, potentially what bioherbicide is best applicable. This way small farmers would have access to shared knowledge regarding weed monitoring, thus essentially creating a smart weed database.
A crucial step is to compose the machine-learning model, that can first classify these weeds. This is what will enable that a weed type and bioherbicide treatment type is returned to the farmer. This is important, because this can serve small farmers by quickly providing data regarding selective bioherbicide application (as well as other herbicide-alternative applications) that is, the appropriate bioherbicide data whichever weed is detected by the machine-learning algorithms.
This platform could help to enhance agricultural sector, as it could contribute to systems, to enable more small farmers to better tend to their crops, or really contribute to economic growth through more efficient farming strategies, while minimising waste of resources, and encouraging the use of optimally distributed natural herbicide alternatives.
Getting a smart weed detection app up in a few minutes:
We can start by getting a weed app up and running in KaggleCloud, for free!