Sanjeet Kumar Verma and Indu Devi*
*Scientist, ICAR-NDRI, Karnal
Introduction
The process of uniquely identifying livestock involves assigning a specific label to each animal and verifying its identity. In modern animal husbandry, this practice is essential for several reasons. It enhances disease prevention and management, ensures product traceability and food safety, boosts breeding efficiency and economic outcomes, and supports the promotion of sustainable practices in animal husbandry. Identifying animals is key for dairy farming operations, including production, feeding, breeding management, handling disease outbreaks, insurance, and trading animals. It enables producers to maintain records of birth rates, production data, health histories, and various other management details. Accurate records equip producers with sufficient information to make informed decisions regarding individual animals or entire herds or flocks. Moreover, it is vital for establishing ownership of a specific animal.
Methods of Identification –
There are various methods for individual identification, which can be classified in the following ways –
- Mechanical (Branding, tagging, tattooing, Ear notching)
- Electrical (RFID)
- Biometric marker (Muzzle print, Iris, Retinal imaging, DNA fingerprinting, etc.)
1.a) Branding – Hot iron was the earliest cattle identification method. The branding irons undergo fire until they become red-hot, and these are applied to the cattle’s hide to kill the cells that grow hair follicles, creating permanent markings. Freeze branding is another type of branding that is similar to hot branding but differs in method; instead of using iron branding, liquid nitrogen or dry ice and alcohol are employed to chill the branding irons.
1.b) Tagging- There are various methods for applying tags; they can be utilized with a neck chain or through piercing. To attach tags to cattle, the ears are pierced between the second and third cartilage ribs, allowing for easy visual identification from both the front and back. While tags are fairly inexpensive and easily readable, the procedure for applying them can be harmful.
1.c) Tattooing- Tattooing is a widely used and lasting technique for identifying cattle by marking a unique combination of permanent letters and numbers onto their skin. As ungulate animals, cattle have their tattoos placed in the ear just above the first cartilage rib to avoid ear tag interference.
1.d) Ear notching- This is a traditionally organised approach commonly used for identifying herds, where the identification of cattle is determined by their birth order within the available breeding, provided it is executed correctly. The litter number, which is often used for piglet identification, is notched onto the right ear of the cattle, referred to as the litter ear, while the individual identification for the cattle is notched on the left ear, known as the individual cattle ear.
2) Radio Frequency Identification (RFID)- This method for identifying animals electronically is particularly suited for ungulates like cattle. RFID consists of a microchip equipped with a small transmitter, radio, and antenna that allows communication with a reader. Various application methods are available for RFID technology, with common forms including microchip implants, ear tags, ruminal boluses, and neck collars. Regardless of the RFID technology applied, a scanner is required to read the microchip, interpret the radio signal as a numerical code, and retrieve the recorded information about the cattle from the herd management software.
Limitations of the above classical animal identification methods-
The application of brands can easily lead to damage or make them hard to read due to improper methods. Cattle movement is restricted when branding, and this process can inflict significant pain from the extremely hot or cold metal. When it comes to tagging, like ear tags, their visual readability can diminish over time. Although the process of placing an ear tag requires little strength and restraint and is generally painless if performed correctly, there is a high risk of infection using this method. Tattooing, a widely used approach for permanent identification in cattle, can be painful and carries the risk of infection if not done in hygienic conditions. The process of ear-notching is relatively painful, and since it relies on visual reading, inaccuracies may occur. Many RFID technologies have a limited range, making them unreliable for monitoring and biometric applications in cattle. Setting up these technologies is costly, and there is a tendency to lose transponders.
3) Biometric features- They are used for identification and must adhere to the requisite standards of uniqueness, stability, and harmlessness to ensure that they are suitable for practical applications. Currently, the biometric features commonly used to distinguish between livestock individuals include retinal vascular patterns, iris patterns, muzzle patterns, facial features, and body patterns.
Advantages of biometric methods- Identification based on facial features does not require any marking or wearing of marking devices on the animal. Visual livestock biometrics have emerged as a highly promising research focus due to their non-invasive nature, with a convenient, fast, and livestock-friendly method.
3.1) Retinal Vascular Patterns-
The arrangement of blood vessels in the retina is a distinctive characteristic that remains unchanged during an animal’s lifetime and can even help differentiate identical twins (Alsaadi, 2015). Consequently, retinal vascular patterns can serve as unique visual identifiers for livestock.
The earlier implementation of livestock identification using retinal vascular patterns was based on the principle of image matching. Although image processing techniques have somewhat mitigated the impact of environmental changes on this matching process, challenges remain. Identifying livestock through retinal patterns is a viable option; however, it comes with difficulties, such as the challenge of acquiring clear retinal images and the effects of external conditions like lighting and flash on image quality. Moreover, obtaining retinal images requires proximity to the livestock, which can induce stress reactions in the animals. Additionally, if the cornea of the livestock’s eye is injured, the system may fail to identify the animal. (Alturk et al., 2019)
Sheep retinal image acquisition (Barron et al., 2008)
3.2) Iris Patterns-
The iris in livestock possesses a structure akin to that of the human iris, characterized by a detailed texture that remains consistent throughout the animal’s life once it matures (Sheng, 2010). Issues such as identifying the iris area and differentiating between the sclera and iris are often associated with iris recognition due to the unique placement of the iris.
Examples of iris images. (He et al., 2008)
3.3) Muzzle Pattern-
The texture pattern on the muzzles of livestock features a dense arrangement of various grooves or bumps, referred to as beads, as well as some flowing structures known as ridges. Muzzle dermatoglyphics resemble human fingerprints and are distinct, serving as a basis for identification (Mishra et al., 1995). In the past, muzzle images were collected by applying ink to the animal’s muzzle and transferring the prints onto paper. With advancements in image acquisition technology, cameras were later used to capture these images.
Gathering images of muzzles is a simpler task compared to collecting retina or iris images. Nonetheless, during the implementation stage, various external factors, such as dirt, sweat around the muzzle area, and lighting conditions, may affect image quality. Furthermore, the movement of livestock can also impact the precision of recognition.
Muzzle image and the unique pattern on the cattle muzzle (Kumar et al., 2017)
3.4) Body Pattern-
The pattern of the body describes the consistent arrangement of hair in various colors on the body of cattle. It is essential to note, however, that not every livestock species has a unique body pattern. Furthermore, certain cattle breeds possess distinct skin colors or markings that can be utilized for identification.
The body pattern image of a dairy cow (Zhang et al., 2023)
3.5) Facial Features-
The facial characteristics of livestock consist of eyes, ears, nose, mouth, facial shape, skin, and hair. These features provide the most immediate visual cues about an individual. Although using facial characteristics for identifying livestock has several benefits, it also has its limitations. For example, various environmental factors like changes in lighting, different angles of imaging, and distances can greatly alter the facial features of livestock. This variation can lead to errors in the identification process. Additionally, the differences in facial features among individuals of the same breed and similar body size in livestock are quite minor.
Challenges of Biometric Methods-
The rapid growth of science and technology, along with the increasing need to modernize the livestock industry, provides a great chance to use computer vision technology for accurately identifying livestock. However, this area faces several challenges, including collecting data, understanding data similarities and differences, and achieving model accuracy and real-time performance.
Challenges in Data Collection- Finding a wide range of high-quality images can be very difficult, especially when dealing with uncooperative animals or working in remote farm areas. Deep learning methodologies frequently require voluminous data for training and learning purposes. Computer vision-based livestock individual recognition requires annotating a substantial number of images to achieve satisfactory performance. However, image annotation requires a lot of time and is a laborious process.
In agriculture, getting good data is challenging because ideal conditions are rarely found on farms. As a result, the data collected often has problems due to various environmental factors. For example, livestock can be hidden by other animals or equipment, and their visibility may change with different lighting conditions.
Keeping livestock in the same posture during data collection can be difficult. This varying posture can cause important details in the images to become blurred, distorted, or missing altogether.
Challenges in Model Accuracy and Generalization-
To tackle the challenges of livestock identification, using complex models is often necessary. These models can extract features and learn more effectively, but they require significant computational power for training and use. Therefore, finding a balance between accurate identification and the complexity of the model is essential. Generalization ability pertains to a model’s capacity to operate effectively on novel, unseen data. Normally, models are evaluated using images that resemble those in the training dataset, and while these models may achieve high performance in evaluations, their performance may drop when faced with images that have different traits.
Conclusions- Reliable systems for identifying animals are essential for data collection and various important management practices. Recognising animals within a herd or flock helps producers and managers make informed decisions based on the records of each animal. Over time, the methods for identifying livestock have significantly advanced. They have transitioned from basic techniques such as branding and ear cutting to the modern, sophisticated computer vision-based identification systems in use today. These identification methods vary, and each comes with its advantages. All methods can be effective when applied appropriately and under suitable conditions. Often, multiple methods are employed together to achieve the highest level of accuracy. With this in mind, it is advisable to assess the needs and anticipated uses of animal identification on a farm before selecting the most suitable identification methods.
References
Alsaadi, I.M. (2015). Physiological biometric authentication systems, advantages, disadvantages and future development: A review. Int. J. Sci. Technol. Res., 4, 285–289.
Alturk, G. & Karakus, F. (2019). Assessment of Retinal Recognition Technology as a Biometric Identification Method in Norduz Sheep. In Proceedings of the 11th International Animal Science Conference, Cappadocia, Turkey, 20–22 October, pp. 20–22.
Barron, U.G., Corkery, G., Barry, B., Butler, F., McDonnell, K. & Ward, S. (2008). Assessment of retinal recognition technology as a biometric method for sheep identification. Comput. Electron. Agric., 60, 156–166.
He, X., Yan, J., Chen, G. & Shi, P. (2008). Contactless autofeedback iris capture design. IEEE Trans. Instrum. Meas., 57, 1369–1375.
Sheng, D.W. (2010). Research on Technology of Cattle’s Iris Recognition. Master’s Thesis, East China Normal University, Shanghai, China.
Mishra, S., Tomer, O. & Kalm, E. (1995). Muzzle dermatoglyphics: A new method to identify bovines. Asian Livest. (FAO), 20, 91–96.
Kumar, S., Singh, S.K. & Singh, A.K. (2017). Muzzle point pattern-based techniques for individual cattle identification. IET Image Process., 11, 805–814. Zhang, R., Ji, J., Zhao, K., Wang, J., Zhang, M., & Wang, M. (2023). A cascaded individual cow identification method based on DeepOtsu and EfficientNet. Agriculture, 13, 279.