Artificial Intelligence (AI)

Cattle Identification Using Muzzle Print Images

August 5, 2023

1719

Cattle Identification Using Muzzle Print Images

Did you know that dogs smell in stereo? Did you know that a dog’s nose print is as individual as a human’s fingerprint? A dog’s nose is an amazing thing. It’s 2023 and people are using their fingerprints to unlock everything from cars to houses to cellphones. This technology, which used to seem futuristic, has now become the norm, and it may be coming to a farm near you.

Fingerprint recognition technology for cellphones was first created in 2011 and is the process of verifying a person’s identity by comparing their fingerprints with previously recorded samples. This recognition is then utilized to unlock certain technology features and can even provide specific data on individuals.

Similar to human fingerprints, cow noses (or muzzles) are detailed, nearly unique, difficult to change, and remain the same over the life of an individual, making them ideal long-term markers of identity. The importance of animal identification has been considered since a long time ago in applications such as cattle classification, cattle tracking from birth to the end of food chain, and understanding animal diseases trajectory and population. Now a robust cattle identification method is an important part for consumers and food industry since the usage of robust cattle identification is related to traceability and registration for breeding and marketing. The muzzle pattern that is correlated with human fingerprints has been considered as a biometric marker for cattle and could be used in identification of bovine animals. We propose to use muzzle photos and applying SURF method to extract and match features between two images. The points extracted from the muzzle pattern images are a good feature for the cattle identification problem especially to handle noisy data. The SURF approach and the proposed matching refinement technique can be a potential method for the beef cattle identification based on the photo image of the muzzle pattern.

Definition

The word Biometrics is derived from the Greek bios (life) and metron (measure). Biometrics has long been used to describe the measurement and logging of biological data, such as animal and human populations (tracking of large numbers of similar life forms). Nowadays, particularly after the increase in human terrorism, biometrics is used to describe methods for non-invasive identification of individuals.

Advantages of biometric techniques

Traditional methods for marking animals can potentially affect their behaviour and cause harm, leading to erroneous research results and poor animal welfare (referanse om merking av pingviner).

Any method used to apply a marker to an animal entails some degree of stress related to capture, handling and restraint. In addition, many common marking procedures also involve tissue damage and therefore cause pain, such as branding (heat, cold or chemicals), tattooing, toe clipping, ear notching and tagging. Furthermore, wearing a mark may alter the animal’s appearance, social interaction, other behaviours and ultimately its survival. An ideal method should identify individuals reliably and permanently with no adverse effects on the animals.

Biometric methods have therefore been developed to recognize animals based on physical characteristics or behavioural signs. Some of these methods have been used for some time for reliable identification of humans.

An animal biometric identifier is any measurable, robust and distinctive physical, anatomical or molecular trait that can be used to uniquely identify or verify the claimed identity of an animal (Barron et al., 2009). Therefore a good biometric trait should be easily presented to a sensor and converted into a quantifiable format, should not subjected to changes over time and should differ in the patterns among the general population, the higher the degree of distinctiveness, the more unique is an identifier.

Biometric methods are non-invasive, do not cause pain and do not alter the appearance of the animal. These methods will therefore have no effect on the behaviour and survivability of the animals, except in cases where repeated capture and/or handling is necessary.

Examples of biometric identification methods

Visual patterns

Some species have external characteristics that are easy to recognize and that are unique for each individual animal. These include colour rings on snakes, body markings of zebras, belly patches in geese and eyespots on the wings of butterflies. These patterns can be photographed or filmed and used to recognize individual animals. Problems may occur in the field in different light settings or surroundings, but new techniques including digital photography and videofilimg have reduced these difficulties. Digital images can also be manipulated to make recognition easier. The method is cheap and at its simplest needs no more than paper and pencil. In addition, observations can be made at a distance, reducing the risk of stress and altered behaviour.

The most obvious biometric marker is the coat pattern of animals which often appears on major body parts as colourations of either fur, feathers, skin or scales. For example, zebras and tigers can be identified from their stripes; cheetahs and African penguins carry unique spot patterns and snakes have coloured rings (Burghardt, 2008).

In a Norwegian study, individuals of the Lesser White-fronted Goose, Anser erythropus, were identified by differences in individual belly patches. The patches were drawn and individuals followed over seven seasons with high accuracy. No individuals were found with similar patches (Øien et al., 1996). Two observers were always present to reduce the risk of mistakes.

Photographic identification has been used since the 1970s to identify aquatic animals such as dolphins and whales (Rugh et al., 1998). Individual bottlenose dolphins can be identified by comparing photographs of their fins, which display curves, notches, nicks and tears. Whales can be distinguished by the callosity patterns on their heads (Wells, 2002).

Nose-prints

This method has been used to identify cattle and was first published by Petersen (1922). The method was developed to avoid the potential for fraud associated with traditional marking methods such as branding, tattooing and ear tags. Both sheep and cattle can be individually identified on the basis of the arrangement and distribution of ridges and valleys on the muzzle (Ebert, 2006).
The method is cheap and simple: ink is applied to the nose and used to make an impression on paper, rather like taking a finger-print from a human. Its accuracy depends, however, on each print being taken in the same way, with the same pressure, ink and type of paper to avoid confusing two animals. They may also be difficult to read due to smearing and they require a trained eye to verify a match. The method is therefore dependent upon the operator’s skills. Nose prints have been shown to be stable over time.

Iris patterns

Iris recognition technology was originally developed for use in humans but has been tested in animals (Musgrave & Cambier, 2002). Iris scanning can be performed rapidly and images can be captured digitally. Its use in animals is limited by the fact that the iris pattern does not stabilise until the animal is several months old and may undergo alteration following injury or infection.

Retinal patterns

The retinal vascular pattern is a unique and distinct biometric trait in animals. It is based upon the branching patterns of the retinal vessels which are present from birth and do not change during the animal’s life. Individual blood vessels in the eye can be detected using a retinal scanner. This pattern can be recorded with a hand-held device about the size of a video camera. Scans from individual animals are registered in a database. Some devices can also measure GPS coordinates. This method can be used when marking cattle and can be compared to nose-prints. The method is also relatively cheap.

Retinal imaging and nose-prints of sheep and cattle were compared by Rusk et al. (1986). Nose-prints are a quicker method than retinal scanning, but retinal scans are easy to analyse for inexperienced operators (Howell et al., 2008). Computer software for the analysis of digital pictures from both retinal scana and nose-prints makes analysis faster, cheaper and more reliable.

Facial recognition

This method has been investigated as an identifier for sheep and was adapted from an independent-components algorithm for human face recognition (Corkery et al., 2007). However, despite the fact that this biometric method has been used by humans for thousands of years, it is difficult to design instruments that can perform facial recognition accurately.

Ear vessel patterns

Inspired by fingerprint identification of humans, the unique blood vessel pattern in the ear of rodents has been studied as a biometric identification method (Cameron et al., 2007). The animal’s ear is photographed from the front while applying backlight to provide a detailed, high-contrast picture of the blood vessels. The branching points of the ear’s blood vessels are automatically detected and compared between two images to identify the individual.

Bite marks

An impression of an animal’s bite marks can be used for identification purposes, in a similar fashion to its use in human forensic medicine. This method is not applicable to all animals and can be difficult to conduct without sedating the individual. For this reason, other methods are preferable.

Saliva sampling

Saliva contains DNA that can be used to recognize individual animals. The method is less invasive than the use of blood sampling to collect DNA.

Movement patterns

It has been suggested that aquatic animals can be identified by analysing their movement patterns using a tri-axial accelerometry device (Shepard et al., 2010). By measuring the movements of animals in three dimensions, their movement patterns can be stored and these can be used to diagnose aberrant behavioural patterns, such as those associated with infections. Accelometery may have the potential to be a powerful tool to produce maps for conservation purposes, where animal movements can be plotted.

Methods of relevance to biometric identification of fish

There are very few published studies where biometric methods have been applied to fish or indeed other aquatic species, for identification purposes. As mentioned above, these have mainly been applied to mammals such as whales and dolphins (Rugh et al., 1998; Wells, 2002). Of the methods described above, external body patterns are likely to be the only biometric methods of any relevance in the foreseeable future:

shapes (e.g. fins, callosity)
patterns (e.g. number and distribution of spots)
colours

Retinal patterns and DNA collection from the skin mucous layer may be of use, but to our knowledge have not been investigated to date.

Animal biometrics is a pattern recognition based system. It is gradually gaining more proliferation due to the diversity of applications and uses in the representing, detecting the visual phenotype appearances, individuals, behaviour analysis, and primary animal biometrics characteristics of animals . Animal biometrics gives a greater impact on recognition techniques for animals or species that are gaining high momentous for the development of innovative computer vision based methodologies for representing, and recognising of species or individuals . In current years, identification of cattle has become extensively used for various applications ranging from the animal registration, traceability, tracking, outbreak, and control of severe disease to behaviour analysis using computer vision and machine learning approaches . However, recognition of cattle has been serious problems for breeding associations in the traditional animal recognition systems throughout the world . It also plays a significant role in the identification, and verification of false insurance claims, missed, swapped, registration of livestock, and the traceability process of cattle . The registration and traceability would stop the efforts for manipulation of animals, trace and follow food, feed, food-producing animal, and substance are supposed to be or expected to be incorporated into a food or feed throughout all stages of production, process, and their distribution . Hence, cattle recognition is essential to control safety policies of animals. It also provides a better management for the food production. Moreover, traceability process of livestock also provides identification of parentage or ownership of animals . The traditional animal recognition methodologies have been classified into several categories, namely (i) permanent identification methodology (PIM), (ii) semi-permanent identification methodology (SIM), and (iii) temporary identification methodology (TIM) . PIM-based technique includes ear-tattoos, the embodiment of microchips, ear-tips or notches, and freeze-branding for the recognition of different cattle. However, PIM methodologies are invasive-based identification methods

Animal biometrics is an emerging research discipline in computer vision, pattern recognition, and cognitive science. It is a promising research field that encourages new development of quantified algorithms and methodologies for representing, detection of visible features, phenotypic appearances of species, individuals and recognition of animals based on their morphological and biometric characteristics . Furthermore, animal biometrics also assists the study of animal trajectory and behaviours analysis of the individual animal or species. Currently, real-world applications of animal biometrics-based recognition system have achieved more proliferation due to a variety of applications and uses, enhancement of quantity and quality of the collection of massive video, captured images of species, collection of ecological data and data processing. However, advance animal biometrics requires better integration of computer vision based methodologies and systems among the scientific disciplines, multidisciplinary researches, ecologists for studies of animal population . Such valuable efforts will be worthwhile due to the enormous perspective of approaches rest with the formal abstraction of biometric characteristics, phenotype appearances, morphological pattern for building well-developed interfaces between different organizational levels of life. Animal biometrics-based recognition system performs on the identification of species or individual animal using extracted features which is similar to recognition of minutiae points in human fingerprints . A feature is defined as a piece of information which is significant for solving the computational task related to a certain application. Moreover, animal biometrics can be applied to genuine understanding of feature representation of animals and classifying the phenotypic appearances of different species or animals based on feature representation of different species. It also identifies the location of the existence of our version with, a recognize the individual behavior as well as to distinguish the morphological image patterns or biometric characteristics of inter-class variation and intra-class of species or individual animal changes over the years .

The SIM approaches have applied for the recognition of animal using an ID-collar and ear-tags. Moreover, the electrical signal based technique, radio frequency identification (RFID), and sketch patterning of the body using paint or dye based techniques have utilised for the recognition of cattle identified as TIM . The embedded ear-tagging in RFID-based recognition techniques are the most characteristic for the identification of individual cattle in herds. It does not require any line of sight visual readings with readers (scanners). The primary drawbacks of RFID-based techniques are not cost-effective, potential losses of RFID transponders, and always need a herd management based software . Therefore, conventional animal recognition methodologies are not satisfactory for the identification of livestock animals . In India, ear-tagging-based techniques suited the extremely expedient for the recognition of different livestock animals . Moreover, in the various countries similar to USA, Australia, Europe, Canada, and Great Britain, embedded RFID in the ear-tags are also applying for the registration, traceability, and recognition of livestock animals . The ear-tagging-based animal recognition techniques have been applied in some ways for the identification of cattle, nevertheless, the significant limitations, and issues of such animal recognition techniques also highlighted in the traditional animal recognition based systems, and livestock framework based systems. Furthermore, the implanted labels of ear-tags, associated with the ear of livestock cattle are also eventually damaged, and damaged due to the long-term usages, and labels of ear-tags can be quickly fraudulent, duplication, and faded with weather conditions . In the direction of cattle recognition, different sketch patterning of the body, and fur of cattle can also be applied to recognise the cattle using a broken colour of different breeds (i.e. Ayrshires, Guernseys, and Holsteins). However, it needs a skilful drawing ability for the colouring process of body surfaces of cattle for getting the better image patterns . The traditional animal recognition methodologies have their boundaries and limitations for recognition of cattle: the nonavailability of efficient, affordable, non-invasive, cost-effective, and scalable animal biometrics based recognition systems for livestock, severe problems of traceability, identification of missed, swapped, false insurance claims, reallocation at slaughter houses of cattle. It also outbreaks dangerous diseases, health management and registration of large population of the animal which are significant problems in the traditional animal recognition systems and livestock framework based system. Therefore, it is a requirement to design and develop an automatic, non-invasive, cost-effective, and robust animal biometric-based-recognition system for identifying individual cattle using muzzle point image pattern. Besides that, all traditional animal recognition techniques, the artificial marking methods (e.g. ear-tips and ear-notches, freezebranding (hot-iron), embedded microchips and RFID) can also be duplicated, fraudulent, and unable to verify the false insurance claims, swapped, and cattle manipulation Due to these significant limitations and failures of the traditional animal recognition based methodologies, livestock framework based systems are explored as better alternative means of cattle recognition. In the available literature, dermatoglyphics of livestock (i.e. ridges, granule, and vibrissae) of muzzle point images are shown. It is different for each breed of cattle. The recognition of muzzle point pattern is very similar to identification of minutiae points in human fingerprint . Accordingly, the muzzle points image pattern of cattle is a suitable, and first animal biometric identifier for the recognition of livestock (especially for livestock), only a few types of research have been done so far and demonstrated that muzzle point image pattern could be used successfully for the identification of individual animals. It gives better solutions to such major problems of previous means of cattle recognition . To address and solve these problems of cattle recognition, we apply the muzzle point image pattern as first animal biometric characteristics for the identification of individual cattle in this paper. Moreover, implemented feature extraction algorithms are motivated by observing that muzzle point (nose print) images have rich skin texture and distinct features such as beads and ridges . The silent sets of extracted texture features of muzzle point image are more discriminate, accurate to recognise the cattle using muzzle point image pattern.

Animal Biometrics Based Recognition System

Animal biometrics-based recognition system is a pattern recognition based system . The recognition system is similar to SLOOP animal identification-based system . The SLOOP identification system is also a pattern retrieval system for identification of species or animal based on their salient set of visual features. It retrieves the information (features) from morphological image pattern and biometric characteristics of species for the recognition of species and individual animal . The SLOOP system uses the cloud computing, machine learning based techniques and crowd sourcing methods to greatly improve the identification accuracy of animals, and tracking movement and analysis of behaviors of animals [84]. Animal biometrics-based recognition system is basically a information retrieval system to extract the discriminatory features of phenotype appearances and visual generic features (e.g., joint stripes of coat pattern of zebra , spot patterning on penguin’s chest, spot points in the tiger body and shark whale , and muzzle point image pattern of cattle for identification of individual animals or species. A phenotype appearance is defined as visual features of species. It is a composition of recognizable characteristics of any organism . These features include discriminatory information of morphological image patterns and biochemical or physiological characteristics of species. The physiological characteristics generally include body structure, body shape, color information, coat patterns, size and specific structural features of organisms .

Component of Animal Biometrics-based Recognition System

Animal biometrics-based recognition system consists of following components-(1) sensors component (for acquisition of data), (2) detection of species based on captured image pattern, and extraction of feature from image pattern, and (3) storage capabilities, (4) similarity matching of query image of species with stored templates in the template database, (5) decision or action executed based on matching scores and defined threshold value, and (6) finally, utilization of extracting features characteristics by the interfacing of various applications or multidisciplinary users and researchers. The acquisition step depicts the capturing of input data from the various study site and data acquisition. The captured images are used for pre-processing of data and perform the measurements and interpretations of pre-processed data for the representation of animals. The animal systems is also applied to study of population and behavior analysis of individual animal by various multidisciplinary researchers, scientists, and engineers. Each of the components of animal biometrics is illustrated to recognize the different species using different marking patterning based on spot point patterns for recognition of individual African penguin.

Iffco Tokyo is digitizing its cattle insurance by deploying an artificial intelligence-powered digital tag developed by Dvara E-Dairy Solutions. TOI had reported in March 2021 how the new biometric technology for livestock could widen the scope of cattle insurance by solving the problem of cattle identity.

Muzzle images of the cattle are collected through the Surabhi mobile application and stored in hi-resolution images as a unique digital identity. Dvara will assist Iffco Tokyo General Insurance Company Limited in digital transformation from the existing cattle identity solutions.

The artificial intelligence-driven mobile application of Dvara E-Dairy captures muzzle images with the mobile phone, compares the cattle’s unique digital identity stored in a secured cloud server, and retrieves the results at the click of a button in less than 60 seconds.

Conventional methods like polyurethane ear tags (PU ear tags) can be easily tampered with and are prone to duplication and fraud. Also, Injectable Radio Frequency Identification ( RFID) tags are expensive and require specialized skills. On the other hand, muzzle printing or nose printing is a unique identifier because it perceives patchy traits on the muzzle of cattle, just like human fingerprints.

The grooves, valleys, and beads structures, cattle muzzle prints have discriminative features, making them unique. Surabhi e-Tag captures these features and stores them in a secure, tamper-proof environment and can significantly improve the penetration of cattle insurance.“Lack of tamper-proof, scalable, unique digital identity of cattle is one of the key reasons for moral hazard, resulting in a higher loss ratio for cattle insurers. Using advanced AI and ML technologies, the critical challenge of improving the quality of images at the time of image capturing is mitigated. We are excited to launch a pilot with IFFCO Tokio General Insurance. ” said Ravi. K.A, Founder & CEO, Dvara E-Dairy Solutions.

Subrata Mondal, EVP – Underwriting, IFFCO Tokio General Insurance, said, “We were exploring a reliable cattle identity process that can be stored digitally and can be accessed anytime. We are happy to partner with Dvara E-Diary to launch Surabhi e-Tag that accurately captures muzzle images.”

According to Dvara, India has the largest cattle population of 300 million. Only 4% of these cattle are insured while 6-8% are funded by lending institutions.

The Times of India : June 29 2021

A huge game-changer

Anand Pejawar, Deputy Managing Director, SBI General Insurance, says that technological advancements in muzzle reading are a huge game-changer for insurers offering cattle insurance. Rural households may suffer financial setbacks in the unforeseen event of loss of cattle and insurance is imperative in such a scenario. However, one of the key challenges faced by insurers is authenticating a claim by correctly identifying the cattle insured, and muzzle reading through a unique digital identity helps address this gap. “We believe that this will potentially grow the market for cattle insurance. Our association with Dvara E-dairy enables us to leverage technology-driven solutions for our cattle insurance offering.”

IFFCO-TOKIO General Insurance Company uses muzzle print as a secondary identification method, along with conventional tags, and is happy with the success ratio of the validation process at the time of claim. This is emerging as a promising technology for identification in bovines, said Subrata Mondal, Executive Vice-President, Head underwriting, Risk Management & Product Development, IFFCO-TOKIO General Insurance Company.

C Soundararajan, Director, Centre for Animal Health Studies, Tamil Nadu Veterinary and Animal Sciences University (Tanuvas), said that muzzle identity technology has been validated at the university through the facilitation of VIF@TANUVAS, an Incubation centre established by the financial support of Entrepreneurship Development and Innovation Institute of Tamil Nadu Government for nurturing start-ups.

Nose print’ ID in Dog

Each dog’s nose is as unique as a human fingerprint, says AI and biometrics company iSciLab.

A South Korean company has developed a biometric recognition tool allowing dogs to be identified by their nose prints.

Once pet owners register the nose pattern and general information of their dog into an app called “Anipuppy”, the information can be easily recalled by scanning the dog’s nose print.

“It’s a 3D biometric algorithm based on AI (artificial intelligence) and deep learning that we have now put into smartphones so that you can take pictures of the nose patterns and use it to identify each animal,” said Sujin Choi, director of iSciLab Corporation.

With the new technology, which the company says is 99.9 per cent accurate, people who find lost dogs can quickly and directly communicate with their owners.