Table of Contents
- 0. Introduction
- 1. Traditional identification technology
- 2. Bar code marking technology
- 3. Radio-frequency identification technology
- 4. Machine vision marking technology
- 5. Molecular Marker Recognition Technology
- 6. Animal individual identification technology research outlook
Individual animal identification plays an important role in breeding, slaughtering, processing, and consumption. In the breeding process, based on the identification of individual pig production performance determination, to guide the selection and breeding of breeding pigs based on the pig individual automatic identification and Internet of Things technology, the collection of pig physiological parameters and breeding information, regulation of the production environment and breeding process, to achieve the goal of precision breeding. In the link of disease prevention and monitoring, the application of marking technology establishes the disease tracing system, records the immunization of individual pigs, quickly traces the source of infection and the transmission pathway when there is an outbreak of disease, discovers the trend of epidemic and the biological pattern of the disease, controls the movement of the source of infection in time, and reduces the economic and social losses. In the product trade and food safety link, establish the labeling and traceability system of pigs and their products to protect consumers from the hazards of food safety accidents. To this end, all countries have concentrated their advantageous scientific research and industrial forces on technological research and have achieved certain results. In this paper, we will focus on the development and evolution of animal individual identification technology and look forward to developing individual identification technology in machine vision and other related fields.
1. Traditional identification technology
Animals were domesticated from the wild state to artificial breeding, and the ancients used various physical methods of identification, such as tattooing or branding of the skin, tail, lips, hooves, and horns, perforation or cutting of the ears, smearing or marking of pigments, wearing collars, anklets or tags, and so on.
1.1 Ear cropping
Ear clipping method is also known as the cut ear method, or ear carving method. With special ear number pliers, in the left and right ears of livestock playing on the notch or round hole, each ear or round hole represents a specific number, ear number composition rules, including digital rules and coding rules. Digital rules are mainly divided into “inside and outside the law” and “up and down the law”, coding rules mainly include “large row method”, “nest row method The coding rules mainly include “big row method”, “nest row method” and “labeling method”.
1.2 Tattoo method
Ear tattoo is one of the conventional methods of permanent identification of animals. Wash and dry the inside of the right ear of the young animal shortly after birth with hot water, and puncture the inside of the right ear with ear tagging pliers. The ear tagging pliers can be placed into different numbered needles, and the puncture is coated with black ink, or soot-alcohol solution, or, for dark-colored or black-eared breeds, a green dye is more appropriate. In addition to ear tattooing, some areas use tattoo hammers to tattoo numbers and mark hogs for slaughter, but this method is now largely obsolete.
The tattoo method must restrain the animal’s head in order to tattoo and identify, which is not suitable for daily use, but the marking accuracy is high, it is not easy to make mistakes, and the incision is small, which is not easy to cause bacterial infection.
1.3 Branding Method
The initial branding method uses the fire branding method, the branding iron is made of mild steel alloy and other excellent thermal conductivity into the shape of characters or symbols, the heated branding iron is placed on the animal’s skin for a few seconds, according to the shape of the body rolling pressure, and then finally cold water or wound oil to relieve skin burns and promote healing. The fire branding method can damage the dermis, affecting the quality of the leather, and the scars of the burns often become infected and pus-filled, resulting in blurred handwriting that is difficult to recognize.
Due to animal welfare reasons, the fire branding method is gradually being replaced by the freeze branding method. The cryo-branding iron is made of copper or bronze alloy and is cooled to -60°C using carbon dioxide cylinders, dry ice-ethanol mixture or liquid nitrogen, which destroys pigment-producing pigment cells in the skin without damaging the hair follicles. The advantages of this method are clear and obvious, extremely easy to recognize, never disappear; easy to operate, less damage to the skin, painless.
1.4 Color method
In the production process, sometimes need to make temporary marking. For example, when giving birth to a litter, it is necessary to use the same color paint to mark out the number of the mother and the litter with the same color paint, you can choose special marking paint, or use easy to wash off the color material for marking.
2. Bar code marking technology
With the development of material and process, wearing marking mainly with plastic ear tag is widely used in global livestock production. Adding barcode marking on the ear tag can realize automatic identification and reading of the ear tag number, on the basis of which, the construction of animal traceability management system and the establishment of perfect food quality and safety system have been used and popularized one after another in various countries.
The United States formed a livestock labeling development group, the development and establishment of livestock labeling work plan, the purpose is to find foreign diseases, can be identified within 48h to determine all the enterprises involved. 2004 began to use with enterprise number or individual identification of the ear tag. Canada has implemented the “Brand Canada” strategy, and in 2008, 80% of agri-food consortium products were traceable to their source. The European Union requires most countries to develop and implement mandatory traceability systems for livestock and meat products, and all products sold after 2002 must have traceability labels. France has implemented mandatory labeling for pigs that are slaughtered. The Netherlands in piglets within a week of weaning plus ear tagging, breeding pigs must be slaughtered or exported at the time of slaughter marking. Denmark on the birth of pigs from the place of birth to the outward transportation, must be marked with ear tags. Australia has a national livestock tagging program that can track livestock from birth to slaughter. Individual livestock are identified with certified ear tags or rumen tags. When a rancher sells a sheep, the tag number must be entered on the application form.
China attaches great importance to the prevention and control of major animal diseases and food quality and safety, and began to implement the animal immunization labeling system in 2001. In 2002, the Ministry of Agriculture issued the “Measures for the Administration of Animal Immunization Labeling”, which stipulates that animal immunization labeling consists of immunization labels and immunization records, and that pigs, cattle, and sheep wear immunization ear tags after being immunized against major epidemics. In 2005, the Livestock Law of the People’s Republic of China was passed, stipulating that animal breeders must mark animals and establish breeding files, and requiring measures to implement the accountability system for animal product quality. 2006, the Ministry of Agriculture promulgated the “Management Measures for Animal Marking and Breeding Files”, and launched the “Animal Marking and Epidemic Disease Traceability System” in three cities and one province, namely, Beijing, Shanghai, Chongqing and Sichuan. In 2006, the Ministry of Agriculture promulgated the Measures for the Administration of Animal Identification and Breeding Records, and launched the pilot construction of the Animal Identification and Epidemic Traceability System in Beijing, Shanghai, Chongqing, Sichuan and one province, and promoted the Animal Identification and Epidemic Traceability System nationwide from 2007. In China, plastic ear tags are used as animal identification. 3.
3. Radio-frequency identification technology
RFID technology is a kind of non-contact automatic identification technology, its basic principle is to use radio frequency signals and spatial coupling transmission characteristics, to realize the automatic identification of the identified object. RFID technology has the characteristics of non-contact identification, high reading rate, strong anti-interference ability, etc., which is applied to the identification of individual animals. There are various ways of RFID identification, such as electronic ear tags, collars, leg rings and other wearable identification, subcutaneous implantable identification, rumen built-in identification, etc.; RFID identification has integrated sensors, signal transceivers and other functions.
Commonly used RFID for individual animals mainly include animal ear tags, collar tags, injectable glass tags and rumen electronic capsules.
Animal ear tag mainly has two types of low-frequency and ultra-high-frequency, the two chips, antennas, packaging methods are different. Low-frequency ear tag RFID chip and antenna composed of intermediary inlay, was placed in the mother tag inside, outside the polyurethane plastic package. The collar type electron label selects the use according to the animal kind and the physiological characteristic, the poultry uses the foot ring and the wing label more, applies in the breeding poultry management, the product traceability and so on; The cow, the sheep and so on uses the collar more, applies in the sub-group system, the automatic feeding system, the milking system and so on. The injection type electronic label uses the special tool to place the electronic label in the animal subcutaneous, through the electronic chip different implantation position effect comparison, found the ear below the position most suitable, the preservation rate is high and not easy to be touched. Livestock injection labeling method will cause animal organism tissue rejection reaction; by the buried position and other factors, there are drop label, broken and in vivo displacement and other problems. The rumen electronic capsule is the electronic tag installed in the acid-resistant cylindrical shell, placed through the animal esophagus into the rumen fluid of ruminants, and generally stays in the rumen of the animal for life. Removal of the injectable electronic tag after slaughter is problematic, and the potential danger of entering the food chain needs to be avoided.
4. Machine vision marking technology
Machine vision is a popular research direction in the field of information technology. The system realizes the positioning, identification and tracking of the target in the scene by analyzing the image or video, analyzing and judging the behavior of the target, and judging the occurrence of abnormalities, and it has been widely used in animal identification in recent years. The non-contact machine vision recognition method can improve the real-time and automation degree of recognition, reduce the cost of breeding management and reduce animal stress, but the recognition accuracy and robustness need to be further improved.
4.1 Nose Recognition Technology
In 1922, Petersen used nose prints to identify individual cattle identity, spraying ink on the nose of the cattle, and then printing it on the paper, using the naked eye to observe and compare the way to match the identity; in 1998, Mishra et al. developed a coding system of cattle through the distribution of nose prints characteristics, together with the ear tags to establish a file on the cattle to realize the whole traceability of cattle products; in 2007, Barry et al. In 2007, Barry et al. collected information about cattle nose prints by photographing the nose prints, used principal component analysis and Euclidean distance classification algorithms to recognize the photos of cattle nose prints, and finally concluded that the accuracy of identifying cattle through nose prints could reach 98.85%. Although the nose pattern has the characteristics of invariance, collectability and high recognition accuracy, but the animal action behavior has the randomness, in the collection of nose pattern pictures and improve the collection efficiency, recognition optimization needs to be further improved.
4.2 Iris Recognition
The iris is the ring-shaped visible part between the pupil and the sclera, which is composed of complex fibrous tissues that stretch with the diameter of the pupil, and the random growth process of the animal before birth causes differences in the organization of the respective iris, with lifelong invariance and variability. The iris in general exhibits an inside-out radial structure containing many interlocking subtle features resembling the shapes of spots, filaments, crowns, stripes, crypts, and so on.
In the infrared light, reflecting its image characteristics of the analog signal is received by the high-resolution camera sampling, after digitization is deposited into the computer, each iris data length of 256 bytes, the entire process in the system is completed instantly. Iris recognition technology has the advantage of high accuracy; and the disadvantage is that the iris technology system cost is too high, need to be more complete imaging conditions and a certain amount of imaging time, the application of more cumbersome in animal breeding.
4.3 Retinal blood vessel identification
Retina is located in the back of the eye is very small nerves, is to feel the light and will be transmitted to the brain through the optic nerve is an important organ, used for biometric identification of blood vessels are distributed in the nerves around the retina, in the collection of retinal data, the scanner sends out a beam of light into the user’s eyes, and reflected back to the scanner, the system will quickly depict the blood vessels of the eye pattern and enter into the database.
Retinal recognition technology has a fairly high reliability, the retinal blood vessel distribution is unique and cannot be forged, even in twin animals, the blood vessel distribution is different; the structural form of the retina remains stable during the growth of the animal; the retinal recognition system has a low misidentification rate, the recording device can obtain 700 feature points from the retina, which makes the misidentification rate of the retinal scanning technology recording device Less than one in a million.
Retinal recognition technology acquisition equipment costs more, the acquisition process is more cumbersome. For retinal scanning devices to acquire retinal images, the distance between the eyes and the recording device should be within half an inch, and the eyes must be at rest while the recording device reads the images, which is not convenient enough to use.
5. Molecular Marker Recognition Technology
Since the establishment of DNA molecular marker technology, a variety of DNA molecular markers have been widely used in the construction of genetic maps, assessment of genetic diversity, as well as individual identification and parentage identification. Commonly used molecular markers for animal individual identification mainly include simple sequence repeats, single nucleotide polymorphisms, etc. Compared with the SSR method, which has many alleles, the SNP method, which has only two alleles, is relatively simple in technology, relatively inexpensive, and beneficial for fully automated analysis. In the traceability system based on molecular markers, it is feasible to carry out individual tracing of livestock, the advantage of which is that it has a very high accuracy rate and significant effect on disease control; the disadvantage of which is that it can only be traced from meat products to farms and breeding animals, and it is not possible to trace the process from the supermarkets to the distributors to the processors to the slaughterhouses.
6. Animal individual identification technology research outlook
With the development of information technology, traditional identification methods no longer meet the requirements of modern farming. Machine vision technology is the fastest growing method with the greatest application potential, based on individual identification, it can realize animal tracking, behavioral perception, physiological monitoring, intelligent control, asset management, “package” solutions, and in the future, it is necessary to further solve the problem of technology research and development and landing in actual breeding conditions; RFID technology is more mature, and is currently replacing manual application more. RFID technology is more mature, is currently a substitute for manual application of more methods, but the need to solve the problem of poor compatibility of the industry standard, and at the same time the development of multi-parameter perception technology based on RFID; biometric identification technology recognition of high accuracy, but the application of the scene, technical equipment and artificial participation in the requirements of the future need to improve the existing hardware and equipment, to improve the accuracy, compatibility, and the development of automation according to the use of the environment and production characteristics of the biometric identification system. .
6.1 Machine Vision Technology
Precision farming mode is increasingly emerging, in full consideration of animal physiological habits and breeding equipment and other factors, animal individual identification will be developed in the direction of contactless, automated, real-time, continuous detection, especially based on machine vision individual identification technology.
In terms of data acquisition technology, the environment of the farm is relatively complex, affected by light and other factors, while the animals cover each other, the acquisition of data will be biased, and it is necessary to develop methods applicable to a variety of farming scenarios for the automatic acquisition of data and target tracking class algorithms used in combination.
In terms of incremental recognition learning models, models trained by applying machine learning methods based on a large number of training samples are only applicable to individual animals in test scenarios. In the case of farms, because of the production needs such as population breeding and group rearing, the situation of individual animals leaving the group or introducing new individuals occurs, and the model needs to be re-trained. There is a need to combine individual recognition models with migration learning to develop incremental recognition models without the need to re-train the models.
In terms of developing embedded devices, although domestic and foreign scholars have achieved good results in the research conducted in the laboratory, the specific applications are still limited. Fusing image data acquisition, feature extraction, and individual recognition together to develop embedded recognition devices has become an urgent problem. The developed embedded devices need to be compatible with deep learning models, and further research is needed to complete deep learning quickly and efficiently in embedded devices.
In terms of intelligent breeding system, based on the animal individual recognition technology, to improve the degree of intelligence of individual information collection, research on how to integrate the information of the cattle’s feed intake, body temperature, activity and body condition score, as well as the impact of the breeding environment on the growth of individual animals, the development of intelligent breeding system.
6.2 RFID technology
In terms of RFID technology for individual animal identification, there are only international standards and national standards for low-frequency identification, while the mainstream technology for domestic large-scale application is UHF identification, which is still blank in terms of technical standards.
First of all, improve the recognition accuracy of UHF marking. Low-frequency identification identification distance is short, low transmission rate, does not have multiple targets at the same time identification performance. UHF RFID reader, can read multiple tags at the same time, reading rate is fast, through the introduction or improvement of the corresponding algorithms can improve the recognition accuracy, to exclude signal interference, applied to large groups of individual animal feeding, egg laying, activities and other behavioral monitoring potential is huge.
Secondly, reduce the working power of UHF marking. The working frequency of UHF marking has 433.92MHz, 862(902) – 928MHz, 2.45GHz and 5.8GHz, etc., in which the animal marking uses more is the first two frequency bands, and most of them are passive marking. There are no generally accepted research results on the long-term effects of electromagnetic radiation from UHF marking on animals, let alone relevant norms and standards. Considering that UHF animal marking is mainly applied in China, domestic scholars should strengthen relevant research.
Finally, to improve the perception of UHF marking ability. the development of RFID perception technology has experienced a gradual transformation from “things” to “people” of the main body. Through further active design and deployment of tags or tag arrays, which will be used in some specific animal behavior perception applications, the tag will be transformed into a lightweight passive sensing device used in various aspects, such as sound perception, vibration perception, etc., to achieve the evolution of the tag from recognition to perception.
6.3 Biometric Identification Technology
Biometric identification of animals includes macro features such as fur pattern, carapace morphology, facial features, and subtle features such as mouth, nose, eyes, and blood vessels. The development of new biometric identification analysis technology and the fusion of two or more biometric identifiers can expand the application of biometric identification and improve the accuracy of the identification system.
In terms of voiceprint recognition technology, Bell Labs proposed the concept of voiceprint for the first time by observing the spectrogram of the voice; scholars from various countries have studied the personality parameters in the voice and proposed the Mel inverted spectral technology and linear predictive analysis technology, which realized a great improvement in the accuracy rate; Gaussian mixture model based on the maximum likelihood probability statistic has become one of the important models for voiceprint recognition due to its advantages of simplicity, reliability, and stability. model. Animal vocal feedback of its health status and physiological growth information will be a hotspot of animal behavior research.
In terms of gait recognition technology, the advantage is its non-contact and long-distance perception. Gait features can be divided into two categories: static feature recognition and dynamic feature recognition, and both static and dynamic features affect the accuracy of gait recognition. Gait recognition for farmed animals is mainly aimed at the occurrence of gait abnormality when the animal is sick, and it is easier to observe the abnormality in the early stage of the disease. The gait recognition technology is still in the research stage, and the recognition accuracy and recognition speed need to be further improved.
Author: Yang Liang