Hybrid features and ensembles of convolution neural networks for weed detection

<span lang="EN-US">Weeds compete with plants for sunlight, nutrients and water. Conventional weed management involves spraying of herbicides to the entire crop which increases cost cultivation, decreasing quality crop, in turn affecting human health. Precise automatic on weeds has been research use. This paper discusses detection using hybrid features is generated by extracting deep from convolutional neural network (CNN) along texture color features. The are extracted moments, gray level co-occurrence matrix (GLCM) Gabor wavelet transform. proposed classified Bayesian optimized support vector machine (BO-SVM) classifier. experimental results read that yield a maximum accuracy 95.83%, higher precision, sensitivity F-score. A performance analysis BO-SVM classifier terms evaluation parameters made images field image dataset.</span>