An Optimized Technique for Plant Identification Through Deep Residual Networks

Abstract

Advancing our knowledge and understanding of the plants around us is very significant and crucial in medical, economic, and sustainable agriculture. Plant image recognition has been an interdisciplinary emphasis in the science of computer vision. Convolutional neural networks (CNN) are used to learn feature representation of 185 classes of leaves, under the benign conditions of rapid advancement in computer vision and deep learning algorithms. A 50-layer deep residual learning framework with 5 steps is built for large-scale plant classification in the natural environment. On the leaf snap data set, the proposed model achieves a recognition rate of 93.09 percent as accuracy of testing, demonstrating that deep learning is a highly promising forestry technology.