Deep learning approach for animal identification / Aya Salama Abdelhady ; Supervised Aly Aly Fahmy , Hisham Ahmed Hassan , Aboulella Otteify Hassanein

Thesis (Ph.D.) - Cairo University - Faculty of Computers and Artifcial Intelligence - Department of Computer Science

Automatic animal identi{uFB01}cation is considered a necessityto guarantee ownership and for decease management and decisions management. The goal of this work is to have an automatic identification system for Arabian horses and sheep without the need for special tools or human intervention. The main contributions in this research lies in the following; for Arabian horse identification; the main contributions lies in discovering special featurefor the eyes of the horse led to using a novel technique in iris segmentation based on Corpora Nigra segmentation, using a novel hybrid approach of deep learning classification based on segmented features for Arabian horse{u2019}s identification, For sheep identification, the main contributions are implementing a novel technique in sheep identification using a hybrid approach of deep learning and optimization, automatic sheep age estimation for the first time through literature using teeth images, and sheep weight estimation using dimension in the image of the sheep body

Issued also as CD