Web-based AI-IoT Multi Classifiers Model of IRIS Images in Real Live Farm Field

Abstract

Combining AI models and IoT devices in the farm and agriculture field would yield the most desired productivity. IoT-enabled cameras significantly help in collecting flowers images in real-time on the farm. To Classify a large dataset of flowers images with multi-species, we need a dedicated and sufficient AI model. The purpose of those captured images is to transfer their features into numerical values like the common IRIS dataset. This research investigates a perfect solution to extract the most relative features from the collected images so that engineers can have IRIS attributes from their original images. Those features represent the IRIS flowers; Sepal width and height, and Petal width and height. The applied methodology is a Conventional Neural Network model with four different architectures; GoogLeNet, VGG-16, AlexNet, and ResNet-50. These models would extract features of the image and then select the most efficient ones. Experiments have proved that SVM has accuracy with 98.89% of classifying the selected features. In the last step, using the comparative technique, nine algorithms were used to classify the IRIS species, which reached 97% of the accuracy.