Data Driven Approach to Leaf Recognition: Logistic Regression for Smart Agriculture

Artificial Intelligence has been involved in the restructuring of farming practices. Compared to conventional methods that involved using pesticides to remove weeds, smart and precision agriculture has increased productivity and produced healthy crops. Plant leaves with varying behaviors should be separated as one of the measures in preserving yield quality. We suggest using logistic regression to implement a four-step AI-based process to classify different leaves. Pre-processing, segmentation and feature extraction, have been used for images classification. Various grapevine and swedish leaves have been used in the work by applying the suggested technique to two different datasets. We used image of size $256\times 256$. During pre-processing, the median filter was used, k-means was used for segmentation, Kaze and Blob were selected for feature extraction. Logistic regression was used for classification. Our findings demonstrate that the suggested strategy beats alternative approaches, yielding an accuracy of 88% on the swedish leave dataset and 83% on the datasets pertaining to grapevine leaves.