Identification of wheat tiller based on AlexNet-feature fusion

Wheat genotype identification is possible with proper recognition of its tiller. It is again challenging to recognize the wheat tiller in complex imaging conditions such as blurred image due to motion and dense appearance due to overlap of heads. The identification of wheat tiller helps to recognize the wheat genotypic and provides knowledge about the variability of growth stages, the orientation of the head, and the presence of awn. This research considered the Global Wheat Head Detection (GWHD) dataset with seven traits of wheat. The multi-feature fusion technique is adapted in AlexNet to enhance the performance of the classifier. The fc6 feature and fc7 feature of AlexNet are concatenated and fed to Linear-SVM to classify the seven traits of wheat and achieved 94.14% of accuracy.