Plant Recognition using Watershed and Convolutional Neural Network

Plant identification by recognizing its leaves has proven to be difficult as many leaves look identical. Hence, the aim of this work is to classify plants based on their leaf structure. The images acquired must undergo necessary image pre-processing like resizing, image enhancement, shadow removal and background removal. When the leaf is surrounded by multiple leaves, the image is segmented by watershed algorithm to separate each leaf. Eight different species of plants can be identified with the help of neural network which is capable of classifying a plant accurately. The final output consists of customized brief yet informative information about the plant.