Evaluation of a 2-D ConvNet for Early and Multiple Target Detection and Binary Classification in Sugarcane Fields Using UAV Images

2D ConvNets are deep convolution-based networks used exclusively for image modality-based computer vision operations. Used widely in image segmentation, object detection, image localization and classification, 2D ConvNets exhibit promising scope in camouflage target localization and detection. In this paper, we have explored LeNet-5 based 2D CNN to detect multiple weed patches of varying sizes using UAV images and successfully classifying them into two classes, viz weed and non-weed.