4 Artificial Intelligence (AI) based Diagnosis

Transcript

Hello!! welcome to the fourth talk of the week that is, Artificial Intelligence (AI) based diagnosis. Plant disease diagnosis has been traditionally been done by visual observation where involvement of experts is an essential criteria and it is always a expensive matter because one has to travel to the field and at certain times finding an important expert is also a difficult for growers in certain regions. So bringing an expert from a far of place it again a time-consuming and expensive process so artificial intelligence is coming up in a way that it is helping to solve certain issues like those. Due to which consulting experts is not a mandatory incurrent context because artificial intelligence is trying to replace some of the human activities through use of certain machine and machine learning processes. The use of technology to replace human activities and guarantee efficiency is known as artificial intelligence. But the question arises can artificial intelligence help improve agricultural productivity?

So we will just see into it what are the technologies in this field has been gaining and how it is helping the growers to identify and detect certain pathogens. So artificial application in agriculture is broadly being used for in terms of (i) agricultural robotics, (ii) soil and crop monitoring and (iii) predictive analytics. Farmers are increasingly using sensors and soil sampling to gather data and this data is stored in farm management systems that allows better processing and analysis. The availability of this data and other related data is paving away to deploy artificial intelligence in agriculture. As a result in number of tech companies is investing in algorithms that are being useful in agriculture. For example image recognition is used in potatoes by AgVoice developed by a Georgia based startup for using natural language toolkit for field notes and yield prediction algorithm based on satellite imagery. Various resources developed several artificial intelligent devices that can identify diseases in plants. For example Tensor Flow is a technique known for transfer learning to teach the artificial intelligence to recognize crop diseases and pests damage. It uses Google’s open source library to build a library of around two thousand 2,756 AI images of cassava leaves from plants in Tanzania. The success was that the AI was able to identify the disease with 98% accuracy. So this shows the effectiveness of artificial intelligence to being deployed in agriculture for detection 0and diagnosis of plant pathogens.

Another technique is using of image segmentation and soft computing technique to detect plant diseases. In image segmentation the process basically is of separating our grouping an image into different parts These parts normally corresponds to something that humans can easily separate and view as individual objects. The segmentation process is based on various features found in the image. This might be color information, boundaries or segments of an image. So this is the process how the image segmentation technique offers first – Image Acquisition takes place then, Image pre-processing is required then, Image Segmentation is done, then Features Extraction in image is done and finally Detection and Classification of plant diseases are obtained. So, these are some of the examples how this technology works. So this is an input and this is the output after image segmentation and based on this input and output images the AI detects the particular disease that is occurring on the plant leaf. Here it is banana leaf disease it is caused by scorch disease. Then the other example is beans leaf which is basically infected by a bacterial leaf spot so, this is the input image and this is the output image and, output image is obtained after the segmentation process is done internally. Similarly, it is used for identification of other diseases such as in rose leaf caused by bacterial leaf spot then other fungal diseases in bean leaf and so on.

iPathology: it is another AI based application that is robotic applications and management of plant and plant diseases in agriculture. So robots application can help in precision plant protection technologies. So, intelligence technologies using machine vision or learning have been developed for plant disease detection and identification. A recognition method based on visible spectrum image processing to detect symptoms of religious like citrus greening which is also named as Huanglongbing (HLB) caused by Candidatus Liberibacter species on citrus leaves. The experimental results showed that detection accuracy is as high as 91.93%. This is again a significant improvement where AI based diagnosis is able to detect more than nine times out of ten times accurately. The huanglongbing detection system can detect the pathogen at pre-symptomatic stage this is again a very significant application of AI based technologies because pre-symptomatic stage is normally not able to traced by human eyes. Similarly, Citrus canker caused by (Xanthomonas axonopodis). It causes foliar symptoms and analyzed to evaluate the efficacy of image analysis. The image analysis was more accurate than visual raters for various symptom types. So robots were also used for image recognition process in case of citrus canker and the symptoms were more accurately diagnosed in comparison to the human eye. Then there are certain mobile apps like for example Plantix which is being widely used in several countries of the world and this is a simple app where even laymen can use for detection and diagnosis of a particular plant disease problem and then and get the information regarding how to manage this problem.

Plantix is a free mobile application which offers farmers and gardeners the possibility to receive decision support directly on their smart phone. Due to image recognition the app is able to identify plant type as well as appearance of a possible disease, pest or nutrient deficiency in the plant. The app can be used very easily by the users where the users have to simply take the image of the infected plant leaf and upload it to the app and then the app recognizes the leaf damage pattern and based on it, it gives an output in terms of information about the probable disease-causing agent and we also provide recommendations for taking adequate control measures for management of that particular problem associated with the plant. So it’s a very simple tool based on artificial intelligence being developed and it can be used by all growers and gardeners which may not have very scientific understanding of the disease problem. So, this is how artificial intelligence is coming in a big way in agriculture system. Particularly it is helping plant disease diagnosis and helping us to understand the causal agent of the plant diseases. So, with this we have come to an end of this today’s talk where we have seen how artificial intelligence is helping us to detect and diagnose a plant disease. In the next talk we will be talking about detection and 0diagnostic challenges for detection of emerging pathogens. So we will see how diagnostics are helping to detect emerging pathogens in various parts of the world. Till then have a good time.

Thank you very much.