An Enhanced Mask R-CNN for Herd Segmentation

Livestock image segmentation is an important task in the field of vision and image processing. Since utilizing the concentration of forage in the grazing area with shielding the surrounding farm plants and crops is necessary for making effective cattle ranch arrangement, there is a need for a segmentation method that can handle multiple objects segmentation. Moreover, the indistinct boundaries and irregular shapes of cattle bodies discourage the application of the existing Mask R-CNN which was primarily modeled for the segmentation of natural images. To address this, an enhanced Mask R-CNN method is proposed for multiple objects instance segmentation to support indistinct boundaries and irregular shapes of cattle body for precision livestock farming. The contributions of this method are in multiple folds: (1) provision of optimal filter size that was smaller than ResNet101 (the backbone of Mask R-CNN) for the extraction of smaller and composite features, thereby, the number of parameters required for the training was decreased; (2) utilization of multiscale semantic features using region proposals and (3) fully connected layer of existing Mask R-CNN integrated with a sub-network for enhanced segmentation. The experiment conducted on pre-processed datasets produced 0.93 mAP higher than any results from existing state-of-the-art methods.