EGYPT-CATTLE MUZZLE Cattle muzzle identifier could lead to cow 'passport'

Record ID: 146883

  • Title: EGYPT-CATTLE MUZZLE Cattle muzzle identifier could lead to cow 'passport'
  • Date: 17th July 2015
  • Summary: (SOUNDBITE) (Arabic) RESEARCHER AND ASSISTANT PROFESSOR, DR HAMDI MAHMOUD, SAYING: "We got the idea from fingerprints, facial scans, vocal scans, iris scans, ear prints. All of this is biometric and we can identify humans through it. Animals also have a biometric print that is located in the muzzle. The muzzle is the area between the two nostrils of the animal."
  • Embargoed: 1st August 2015 13:00
  • Keywords:
  • Location: Egypt
  • Country: Egypt
  • Topics: General
  • Reuters ID: LVA7UJYZPW46HC7LXU9OUJVG112K
  • Aspect Ratio: 16:9
  • Story Text: An Egyptian university professor has developed a computerised method of classifying and identifying cattle through their muzzle - the front of the animal's nose and mouth.
    Hamdi Mahmoud is assistant professor at Cairo's Beni Suef University. Along with colleague Hagar Reda, he has developed a machine learning system known as a multi-class support vector machine (MSVM) that identifies cow muzzle prints in order to properly identify each animal. He says it is 94 percent accurate.
    Each cow muzzle has specific ridges and valleys between their nostrils that are distinctive and unique.
    "We got the idea from fingerprints, facial scans, vocal scans, iris scans, ear prints. All of this is biometric and we can identify humans through it. Animals also have a biometric print that is located in the muzzle. The muzzle is the area between the two nostrils of the animal," said Mahmoud.
    In their research paper Mahmoud and Reda argued that cattle muzzle classification can be considered as a biometric identifier.
    In collaboration with Reda, Mahmoud photographed the unique muzzle print of 100 cows and created a database.
    The professor described the software as part of an 'Artificial Neural Network' which can be taught the digital signature of each muzzle through a set of unique numbers.
    Twelve photographs of each animal are used to create the digital signature. Another eight are used to teach the computer to identify the signature.
    "We gathered a database where we took photographs of 100 animals," said Mahmoud. "Each animal had twenty photographs taken of it, so we are talking about a database of almost 2,000 photographs. We built a system or a computer program with a certain technique where we can identify the animal. We have a camera which photographs the muzzle afterwards and then the animal is identified."
    Until now livestock have commonly been identified through methods like ear tags, iron markings, or tattoos.
    Veterinarians have been looking for alternative means of identifying animals without inflicting harm on them.
    "Why did the veterinarians come to us? Firstly, because the animal had to be physically harmed to be identified," said Mahmoud. "If we had an animal stock of 2,000 animals, and they had to be vaccinated, how will you identify each animal? Can you imagine if you had 2,000 animals consisting of our very nice black buffalos who were all close to the same age and size, but all look similar? How can you identify them? The older methods consisted of piercing the ear or ironing and tattooing it using liquid nitrogen or iron. So they now are looking for humane methods of identification such as identifying humans through their eyes and fingerprints using devices."
    Mahmoud said he hopes a future device could "issue a birth certificate" for every animal.
    "We should manufacture devices. There should be a device with a camera which is present in every horse farm and every animal farm," he said. "Consequently every animal is born and is subject to enrolment similar to when you give birth to your child and you issue a birth certificate. We take four or five photographs of the cattle which is used in all livestock animal farms. We will then have an advanced and clean method of identifying animals."
    Mahmoud and Reda have received a number of international offers to develop the system further. They recently published a paper on the research in the International Journal of Image Mining.
  • Copyright Holder: REUTERS
  • Copyright Notice: (c) Copyright Thomson Reuters 2015. Open For Restrictions - http://about.reuters.com/fulllegal.asp
  • Usage Terms/Restrictions: None
Register