Enhanced Mask R-CNN for herd segmentation

Livestock image segmentation is an important task in the field of vision and processing. Since utilizing concentration forage grazing area with shielding surrounding farm plants crops necessary for making effective cattle ranch arrangements, there a need method that can handle multiple objects segmentation. Moreover, indistinct boundaries irregular shapes bodies discourage application existing Mask Region-based Convolutional Neural Network (Mask R-CNN) which was primarily modeled natural images. To address this, enhanced R-CNN model proposed instance to support precision livestock farming. The contributions this are folds: 1) optimal filter size smaller than residual network extracting composite features; 2) region proposals multiscale semantic 3) R-CNN’s fully connected layer integrated sub-network experiment conducted on pre-processed datasets produced mean average (mAP) 0.93, higher results from state-of-the-art models. Keywords: farming, segmentation, mask R-CNN, herd, enhancement DOI: 10.25165/j.ijabe.20211404.6398 Citation: Bello R W, Mohamed A S A, Talib Z. Enhanced herd Int J Agric & Biol Eng, 2021; 14(4): 238–244.