Skip to main content
Fig. 3 | Plant Methods

Fig. 3

From: SpikeSegNet-a deep learning approach utilizing encoder-decoder network with hourglass for spike segmentation and counting in wheat plant from visual imaging

Fig. 3

Flow diagram of SpikeSegNet: The network is developed for pixel-wise segmentation of objects (or spikes) from the wheat plant. SpikeSegNet is a combination of two proposed feature network namely Local Patch extraction Network (LPNet) and Global Mask refinement. Network (GMRNet). a The visual image of size 1656 * 1356 is divided into patches (b) of size 256 * 256 and fed into the LPNet network to extract contextual and spatial features at local patch level. Output of LPNet is segmented mask image patches (c) of size 256 * 256 which are then combined (mergeLPmask) to generate the original mask image of size 1656 * 1356 (d); mergeLPmask image may contain some sort of inaccurate segmentation of the object (or, spikes) and are refined at global level using GMRNet network; before passing through GMRNet, it is resized to 256 * 256 (e) to reduce the network complexity. The output of GMRNet network is nothing but the refined mask image (f) containing spike regions only

Back to article page

Plant Methods

ISSN: 1746-4811