Automated plant disease diagnosis using innovative android App (Plantix) for farmers in Indian state of Andhra Pradesh
Srikanth Rupavatharam: International Crop Research Institute for Semi-Arid Tropics
<div>Agricultural extension systems are stretched beyond physical limits while assisting farmers during crop production in the developing world. Information and Communication Technologies are booming worldwide with the introduction of mobile applications in smart phones. Progressive Environment Agriculture Technologies (PEAT), a German startup company developed a mobile application ‘Plantix’ that assists in detecting damage on plants with the help of a smart phone image. Mobile internet connectivity enables uploading of image (time and geo-tagged) to a cloud based server where machine learning based algorithms are used to process the optical patterns and detect the diseases. This information is conveyed back to the user within few seconds along with suggested knowledge on disease management including biological and chemical control methods. Plantix is presently downloaded by more than 0.520 million users and has an image data base of over 2 million plant damages. Plantix is trained on more than 250 plant damages including diseases, pests and nutrient disorders. Accuracy in identification relies on building a robust data base for every disease through image tagging performed by subject matter experts. For over 100 classes the detection accuracy of plantix is over 90 %, however in some cases like early and late leaf spots of peanut an accuracy of over 95 % (15,000 images) is witnessed. It has been observed that accuracy of detection depends upon quality of data set that increases with rise in number of independent images. There is potential in enhancing detection accuracy and include more plant damages and then the challenge to decipher disease complexes.</div>
View Presentation |
