Segmentation of Rice Seedling using Deep Learning Algorithm
Avinash Kumar, Ashish Rajanand, Anurag Deep Kujur, Yogesh Kumar Rathore, Rekh Ram Janghel
Abstract
The objective of deep learning is used to recognise crops and weeds on the farm. It can be also termed as smart farming. To reduce the pollution caused by spraying excess fertilizer and herbicide in the rice crops, it becomes necessary to correctly locate the rice seedling and weed. Rice seedling Segmentation remains a difficult process. In this paper, we propose a method called semantic segmentation using FCN, Unet, Fast-SCNN, SegNet to differentiate weed from crops. The average accuracy of SegNet is 92.08% and using Unet is 87.5% which is an evolved model of CNN. We have illustrated different algorithms for this process but the main architecture is SegNet which has shown best accuracy among them.