System and method for individual animal identification

The animal identification system uses facial recognition to uniquely identify and track animals, addressing the limitations of existing methods by providing a cost-effective and versatile solution for animal tracking across different producers and processors.

CN119072710APending Publication Date: 2024-12-03TRUETRACE BIOMETRICS INC
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202380035460.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2022-04-22
Filing Date
2023-04-10
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

Existing methods for uniquely identifying and tracking individual animals across different producers and processors are non-standard, costly, and lack universality, particularly with ear tags and RFID tags.

Method used

An animal identification system utilizing facial recognition technology, including an image capture device, communication channel, and computer with a processor, to capture and process animal images, generate data vectors, and compare them for identification, eliminating the need for physical tags or chips.

Benefits of technology

Provides a unique, cost-effective, and versatile method for identifying and tracking animals over time and across geographic regions, leveraging their distinct facial features for recognition, enhancing herd management and productivity assessment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119072710A_ABST
    Figure CN119072710A_ABST
Patent Text Reader

Abstract

An animal recognition system accepts images, determines whether animal faces are present in the images, and then translates these animal faces to allow comparison with known animal faces. The comparison process uses vectorized data to determine similarities between known animal faces and unknown animal faces, in some embodiments, using distance or difference functions calculated on or among the vectorized data. The system allows known animal faces to be captured and added to known animal faces stored in the system without retraining the comparison algorithm. The system is improved in part over time according to user feedback.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] Related Applications

[0002] This application claims priority to and the benefit of U.S. Provisional Patent Application No. 63 / 333,812, filed on April 22, 2022, the entire contents of which are incorporated herein by reference for all purposes. Background Art

[0003] Individuals in the business of raising and processing animals desire more efficient ways to uniquely identify and track individual animals over time. Existing methods for performing such processes involve ear tags that are non-standard across producers and processors and therefore not unique in identifying the animal. Other methods include embedding microchips or RFID tags, but these methods are costly and are not unique across producers and processors. Summary of the invention

[0004] In one aspect, the present disclosure provides an animal identification method, which includes an animal registration phase, which includes: using a device to capture at least one image of a first known animal; using a model to determine a first data vector based on at least one image of the first known animal; receiving a first identification information group of the first known animal on the device; and saving the first data vector and the first identification information group of the first known animal to a database. The animal registration phase also includes using the device to capture at least one image of a second known animal; using the model to determine a second data vector based on at least one image of the second known animal; receiving a second identification information group of the second known animal on the device; and saving the second data vector and the second identification information group of the second known animal to a database. The animal identification method also includes an animal recognition phase, which includes: using the device to capture at least one image of an unrecognized animal; using the model to determine a new data vector based on at least one image of the unrecognized animal; comparing the new data vector with the first data vector and the second data vector and identifying the unrecognized animal as the first known animal or the second known animal; and displaying the recognized animal including the corresponding identification information group on the device.

[0005] In some embodiments, capturing at least one image of the first known animal includes capturing a first image of the first known animal at a first perspective and capturing a second image of the first known animal at a second perspective.

[0006] In some embodiments, capturing at least one image of the first known animal includes capturing a video of the first known animal.

[0007] In some embodiments, the animal identification phase further includes receiving feedback at the device regarding the accuracy of identifying the unidentified animal as the identified animal.

[0008] In some embodiments, the animal is a cow, sheep, horse, pig, goat, chicken, dog, or cat.

[0009] In some embodiments, the first set of identification information includes: ear tag, batch ID, sex, notes, feeding performance, lameness, disease, antibiotic status, pen movement, or any combination thereof.

[0010] In some embodiments, the model is a machine learning model.

[0011] In some embodiments, the model is trained using an augmented dataset.

[0012] In some embodiments, the augmented data set includes: rotating images, scaling images, flipping images, brightness adjusting images, generating images, or any combination thereof.

[0013] The present disclosure provides, in one aspect, an individual animal identification system, the individual animal identification system comprising: an image capture device capable of capturing at least one digital image; a communication channel connected to the image capture device; and a computer connected to the communication channel. The computer comprises at least one processor and a non-transitory computer-readable memory. The image capture device captures at least one digital image and sends the at least one digital image to the computer using the communication channel so that the computer: detects the presence of an animal face in at least one digital image sent from the image capture device; determines and stores the coordinates of the position of the face of the animal detected in at least one digital image; applies at least one transformation of the detected animal face in at least one digital image to create at least one transformed digital image; compares at least one transformed digital image with at least one stored digital image using concurrent vector comparison to calculate a similarity between at least one transformed digital image and at least one stored digital image; sends the calculated similarity to the image capture device using the communication channel; receives feedback on the accuracy of the calculated similarity between at least one transformed digital image and at least one stored digital image; and updates the concurrent vector comparison calculation to improve accuracy.

[0014] In some embodiments, the image capture device captures data including the three-dimensional orientation of objects in the digital image.

[0015] In some embodiments, the image capture device captures data including the depth of objects in the digital image.

[0016] In some embodiments, detecting the presence of an animal face in at least one digital image sent from an image capture device includes detecting key points in the image, including the animal's eyes, nose, ears, mouth, and other known facial features.

[0017] In some embodiments, detecting the presence of an animal face in at least one digital image sent from the image capture device includes detecting a depth of the animal.

[0018] In some embodiments, concurrent vector comparison uses a vector of values ​​to simultaneously compare multiple values ​​in different vectors, which can consist of a triplet loss technique in which a reference input image is compared with a matching input image, and a non-matching input image to minimize the difference between the reference input image and the matching input image and maximize the distance between the reference input image and the non-matching input image.

[0019] In some embodiments, at least one transformation of a detected animal face in a digital image includes at least one of: cropping to eliminate non-facial data, orienting the detected animal face for standard analysis and comparison, scaling the image of the detected animal face to a standard size, detecting standard facial features (such as the distance between eyes), and deriving at least one vector of image data from the detected animal face.

[0020] In some embodiments, the at least one digital image may be a series of digital images including a video image.

[0021] In some embodiments, the concurrent vector comparison utilizes a comparison algorithm to calculate differences between different animal faces, and using the difference calculation may select a stored image that is similar to at least one transmitted image or may indicate that no similar image exists in the non-transitory memory.

[0022] Other aspects of the disclosure will become apparent by consideration of the detailed description and accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The above aspects and other features of the animal identification system are explained in the following description herein in conjunction with the accompanying drawings.

[0024] Figure 1 is a schematic diagram showing an animal identification system for individual animal identification.

[0025] Figure 2 is a flow chart illustrating a method of registering known animals.

[0026] Figure 3 is a flow chart illustrating a method of identifying an unidentified animal.

[0027] Figure 4 is an image of an animal showing detection of the animal's face and cropping of data outside the shown box.

[0028] Figure 5Examples of comparison data vectors generated and used to determine positive matches and negative matches are shown.

[0029] Figure 6 It is shown how feedback can be used to improve the comparison criteria to more clearly distinguish matching images from non-matching images.

[0030] Figure 7 is a flow chart of a method for adding known animals to a database.

[0031] Figure 8 is a flow chart of a method for identifying an unidentified animal.

[0032] Fig. 9 is a flow chart of a method including an animal registration phase and an animal identification phase.

[0033] Before any embodiments are explained in detail, it should be understood that the invention is not limited in its application to the details of construction and the arrangement of components set forth in the following description or illustrated in the following drawings. The invention is capable of other embodiments and of being practiced or carried out in many different ways. DETAILED DESCRIPTION

[0034] The section headings used in this section and the entire disclosure herein are for organizational purposes only and are not intended to be limiting.

[0035] 1. Definition

[0036] Unless otherwise defined, all technical and scientific terms used herein have the same meanings as those of ordinary skill in the art are generally understood. In the event of a conflict, the present document (including definitions) shall prevail. Preferred methods and materials are described below, although methods and materials similar or equivalent to the methods and materials described herein may be used in the practice or testing of the present disclosure. The entire contents of all publications, patent applications, patents, and other references mentioned herein are incorporated herein by reference. The materials, methods, and examples disclosed herein are merely illustrative and are not intended to be limiting.

[0037] As used herein, the terms "comprises," "including," "having," "having," "may," "including," and variations thereof are intended as open transitional phrases, terms, or words that do not exclude the possibility of additional actions or structures. Unless the context clearly indicates otherwise, the singular forms of "a," "and," and "the" include plural referents. The present disclosure also contemplates other embodiments of the embodiments or elements presented herein that "comprise," "consist of," and "consist essentially of," whether or not explicitly stated. Furthermore, unless the context requires otherwise, singular terms shall include the plural and plural terms shall include the singular.

[0038] For the expression of numerical ranges herein, each intermediate number with the same precision is explicitly considered. For example, for the range of 6 to 9, in addition to 6 and 9, the numbers 7 and 8 are also considered, and for the range of 6.0 to 7.0, the numbers 6.0, 6.1, 6.2, 6.3, 6.4, 6.5, 6.6, 6.7, 6.8, 6.9 and 7.0 are explicitly considered. Unless otherwise specified herein, the description of the range of values ​​herein is intended only to be used as a shorthand method for individually referring to each independent value falling within the range, and each independent value is incorporated into the specification as described individually in this article. For example, if the concentration range is specified as 1% to 50%, it is intended to explicitly list values ​​such as 2% to 40%, 10% to 30%, or 1% to 3% in this specification. These are only examples of the contents specifically intended to be disclosed, and all possible combinations of numerical values ​​between the enumerated minimum and maximum values ​​and including the minimum and maximum values ​​should be considered to be explicitly stated in the present disclosure.

[0039] As used herein, the terms "processor" and "central processing unit" or "CPU" are used interchangeably and refer to a computer capable of reading a program from a computer memory (e.g., a ROM or other computer memory) and executing a set of steps according to the program. As used herein, the term "processor" (e.g., a microprocessor, a microcontroller, a processing unit, or other suitable programmable device) may include, among other things, a control unit, an arithmetic logic unit ("ALC"), and a plurality of registers, and may be implemented using known computer architectures (e.g., a modified Harvard architecture, a von Neumann architecture, etc.). In some embodiments, the processor is a microprocessor that may be configured to communicate in an independent and / or distributed environment, and may be configured to communicate with other processors via wired or wireless communications, wherein such one or more processors may be configured to operate on one or more processor-controlled devices, which may be similar or different devices.

[0040] As used herein, the term "memory" is any memory storage device and is a non-transitory computer-readable medium. The memory may include, for example, a program storage area and a data storage area. The program storage area and the data storage area may include a combination of different types of memory, such as ROM, RAM (e.g., DRAM, SDRAM, etc.), EEPROM, flash memory, hard disk, SD card or other suitable magnetic storage device, optical storage device, physical storage device or electronic storage device. The processor can be connected to the memory and execute software instructions, which can be stored in the RAM of the memory (e.g., during execution), the ROM of the memory (e.g., on a generally permanent basis) or another non-transitory computer-readable medium such as another memory or disk. In some embodiments, the memory includes one or more processor-readable and accessible memory elements and / or components, which may be located inside a processor-controlled device, outside a processor-controlled device, and may be accessed via a wired or wireless network. The software included in the implementation of the method disclosed herein may be stored in the memory. The software includes, for example, firmware, one or more applications, program data, filters, rules, one or more program modules and other executable instructions. For example, the processor may be configured to retrieve from the memory and in addition execute instructions related to the process and method described herein.

[0041] As used herein, the term "computer-readable medium" refers to any device or system for storing information (e.g., data and instructions) and providing it to a computer processor. Examples of computer-readable media include, but are not limited to, DVDs, CDs, hard drives, magnetic tapes, and servers for streaming media over a network, whether local or remote (e.g., cloud-based).

[0042] "About" and "approximately" are used to provide flexibility to numerical range endpoints by providing that a given value can be "a little above" or "a little below" the endpoint without affecting the desired result.

[0043] As used herein, the term "coupled" is defined as "connected," although not necessarily directly and not necessarily mechanically. The term coupled should be understood to mean physically, magnetically, chemically, fluidically, electrically or otherwise coupled, connected or linked, and does not exclude the presence of intermediate elements between coupled elements, unless there is specific language to the contrary.

[0044] As used herein, the term "in electronic communication" refers to electronic devices (e.g., computers, processors, etc.) that are configured to communicate with each other by direct or indirect signaling. Likewise, a computer that is configured to send information to another computer or device (e.g., by cable, wire, infrared signal, telephone line, radio waves, etc.) is in electronic communication with another computer or device. As used herein, the term "send" refers to moving information (e.g., data) from one location to another (e.g., from one device to another) using any suitable means.

[0045] As used herein, the term "network" generally refers to any suitable electronic network, including but not limited to a wide area network ("WAN") (e.g., a TCP / IP-based network), a local area network ("LAN"), a neighborhood area network ("NAN"), a home area network ("HAN"), or a personal area network ("PAN"), which employs any of a variety of communication protocols, such as Wi-Fi, Bluetooth, ZigBee, etc. In some embodiments, the network is a cellular network, such as a Global System for Mobile Communications ("GSM") network, a General Packet Radio Service ("GPRS") network, an Evolution-Data Optimized ("EV-DO") network, a GSM Enhanced Data Rates for Evolution ("EDGE") network, a 3GSM network, a 4GSM network, a 5G New Radio, a Digital Enhanced Cordless Telecommunications ("DECT") network, a Digital AMPS ("IS-136 / TDMA") network, or an Integrated Digital Enhanced Network ("iDEN") network, etc.

[0046] As used herein, the term "animal" refers to any animal (e.g., mammals, reptiles, birds, amphibians, fish), including but not limited to beef cattle, dairy cows, horses, pigs, sheep, goats, chickens, ducks, turkeys, cows, dogs, cats, non-human primates, livestock, horses, etc.

[0047] As used herein, the term "known animal" refers to an animal that is known to a user or recognizable by a user, such that the user can provide further information about the animal.

[0048] 2. Animal identification system

[0049] The animal identification system described herein provides a unique and inexpensive system and method for capturing data about animals, identifying the animals through the data, and then identifying and tracking the animals over time, in different locations and under different conditions. As all humans have unique fingerprints, humans and animals also have unique facial features that, once captured, can be used to identify individuals. Without the need to attach, remove, and reattach new ear tags, or to insert microchips or RFID chips at great time and cost, facial recognition can be quickly captured, stored electronically, and used by producers and processors across geographic regions and over time. In addition, animals are not subject to the privacy restrictions associated with humans, making the recognition technology more powerful, more versatile, and more open than human facial recognition for use by multiple parties.

[0050] refer to Figure 1 , individual animal identification system 10 includes device 14, network 18 including communication channel 22, and computer 26. In some embodiments, communication channel 22 is wired. In other embodiments, communication channel 22 is wireless.

[0051] In the illustrated embodiment, device 14 is an image capture device capable of capturing at least one digital image. In some embodiments, device 14 can capture video. In some embodiments, device 14 can capture data including the three-dimensional orientation of objects in the digital image. In some embodiments, device 14 can capture data including the depth of objects in the digital image. In some embodiments, device 14 is a user device, such as a cell phone, tablet computer, personal computer, or any other suitable user device. In the illustrated embodiment, user device 14 is a cell phone with a display.

[0052] In the illustrated embodiment, the device 14 is located on or near the user in a first location 30. In some embodiments, the first location 30 is anywhere the user is located and moves with the user. As explained in more detail herein, in some embodiments, the device 14 is configured to capture an image of an animal 34 at the first location 30.

[0053] Continue to refer Figure 1 , communication channel 22 is connected to image capture device 14. In some embodiments, network 18 is a cloud-based network connected to the Internet. In some embodiments, network 18 includes multiple databases 38 for storing various data disclosed herein. In some embodiments, data collected from image capture device 14 is stored in database 38. In some embodiments, database 38 includes information collected from multiple users.

[0054] In the illustrated embodiment, data on device 14 is sent to network 18 via communication channel 22 (e.g., via WiFi) and stored in database 38. In some embodiments, data flows bidirectionally between device 14 and network 18. In other words, device 14 can send data to network 18, and network 18 can send data to device 14. In the illustrated embodiment, image capture device 14 captures at least one digital image and sends the at least one digital image to computer 26 using communication channel 22.

[0055] In some embodiments, the computer 26 is a mobile phone, tablet computer, etc. connected to the network 18. In the illustrated embodiment, the computer 26 is connected to the communication channel 22. The computer 26 includes at least one processor and a non-transitory computer-readable memory. In some embodiments, the computer 26 can be any suitable network-connected device that is capable of accessing and manipulating the database 38. In the illustrated embodiment, the computer 26 is located at a second location 42. In some embodiments, the second location 42 is spaced apart from or remote from the first location 30. For example, the first location 30 can be a ranch or a feedlot, and the second location 42 can be an office or a server room. Both the first location 30 and the second location 42 are connected to the network 18.

[0056] As further detailed herein, the animal identification system 10 is a system in which an image capture device 14 (e.g., a phone, a digital camera with wifi communication capabilities, a tablet, etc.) is used to capture at least one image of an animal or, in some embodiments, a video of an animal. The image capture device 14 sends the at least one image across a communication channel 22 (e.g., wired, Bluetooth, or WIFI) to an internet connection, where the communication channel continues across the internet to a computer 26 having at least one processor and non-transitory memory (such as a hard drive), where the at least one image is stored, processed, and utilized. The system 10 provides integrated data storage, computation, and comparison of animal images, which can be universally used for all animal types to identify animals. As further detailed herein, the registration and identification of animals by the system 10 provides a means for herd management, health monitoring, and productivity assessment.

[0057] 3. Data processing and expansion

[0058] The data collected by the device 14 may include an image, multiple images, video, user input data, or any combination thereof. The data from the device 14 is sent to a database 38 on the network 18 via the communication channel 22. The stored data may then be processed, transformed, and augmented as disclosed herein.

[0059] In some embodiments, computer 26 uses stored instructions to perform one or more transformations on one or more images sent from the image capture device. In some embodiments, such transformations include detecting animal faces, cropping non-facial data from at least one image to eliminate extraneous data, reorienting animal faces detected in at least one image to a standard location within the image, resizing faces to normalize calculations and feature detection, dithering and smoothing, or other such image transformations.

[0060] In one example, computer 26 executes stored instructions to detect the presence of an animal face in a digital image sent from image capture device 14. This processing may be performed regardless of whether the animal face is a known animal face or an unrecognized animal face. Figure 4 , an animal face is detected in the image, and a region containing the animal face is delineated using a bounding box 46. Portions of the image outside of the bounding box 46 may be cropped or eliminated for further processing, conversion, or comparison. In some embodiments, detecting the presence of an animal face in a digital image from device 14 includes detecting key points in the image, including but not limited to the animal's eyes, muzzle, ears, mouth, or other known facial features. In some embodiments, detecting the presence of an animal face in a digital image from device 14 includes detecting the depth of the animal.

[0061] In one example, computer 26 executes stored instructions to determine and store coordinates of the location of a detected animal's face in a digital image.

[0062] In one example, computer 26 executes stored instructions to apply at least one transformation of the detected animal face in the digital image to form at least one transformed digital image.

[0063] In some embodiments, computer 26 uses stored instructions to perform one or more calculations on the converted image, including in some embodiments generating one or more data vectors and calculation data from the transmitted image, which can be used to compare with one or more data vectors and calculation data from the image stored in the non-transitory component. As further detailed herein, in some embodiments, this comparison can use a comparison technique in which a known image is compared to one or more transmitted images to determine a match, which can be a triple-derived comparison, a neural network, a heuristic method, or a random or deterministic model and other artificial intelligence techniques. The comparison technique can generate one or more differences that can be used to determine the similarity between two or more images, or indicate that there is no such similarity between the images.

[0064] In some embodiments, transformation of a detected animal face in a digital image includes at least one of cropping to eliminate non-facial data, orienting the detected animal face for standard analysis and comparison, scaling the image of the detected animal face to a standard size, detecting standard facial features (such as the distance between the eyes), and deriving at least one vector of image data from the detected animal face.

[0065] In some embodiments, image data received from device 14 is augmented to improve system 10 performance. In some embodiments, a generative model is used for data augmentation. In some embodiments, a generative model creates synthetic data and is applied to generate training data. Data augmentation is important and helpful when there is limited training data. For example, unlike images of human faces where sufficient training data can be obtained from various sources (e.g., the Internet), images of beef and dairy cow faces have limited training data. The data augmentation disclosed herein advantageously enhances the training data available for training various models on various animal faces. In some embodiments, a generative model creates synthetic data, which is then applied as training data for tasks such as bounding box annotation, segmentation, and animal face recognition.

[0066] Data augmentation is a technique used to improve the performance of models (e.g., machine learning models), especially in image-based tasks with limited training data. Data augmentation is a technique to artificially increase the size and diversity of a dataset by applying various transformations to the original data. In image-based tasks, data augmentation techniques include rotation, scaling, flipping, brightness adjustment, and other suitable image transformations. Data augmentation improves the generalization ability of the model by providing more diverse training examples, thereby reducing the overfitting problem.

[0067] A generative model is a class of machine learning models that can generate new data instances by learning the underlying distribution of input data. In other words, a generative model can create synthetic data for data augmentation. In one embodiment, the generative model is a generative adversarial network (GAN). GAN includes two neural networks, a generator and a discriminator, which are trained simultaneously. The generator learns to create realistic synthetic images, while the discriminator learns to distinguish between real images and generated images. As another example, the generative model is a variational autoencoder (VAE), which is a class of autoencoders that learn to encode input data into a low-dimensional latent space and then decode it back to the original data form. VAE is trained to learn the distribution of input data in the latent space, enabling it to generate new data instances.

[0068] In one embodiment, generative models are used to improve bounding box annotation tasks. For example, GANs and VAEs can be used to generate synthetic images of beef cattle with associated bounding box annotations. By training a generative model on images of beef cattle with their corresponding bounding boxes, the model can learn to generate new images with realistic object positions and sizes. The synthetic images and their associated bounding boxes can then be used to augment the training dataset, thereby improving the performance of facial recognition models of beef cattle faces.

[0069] In one embodiment, the generative model is used to improve the segmentation task. For example, the generative model can be applied to create synthetic images with corresponding segmentation masks. For example, by training the model on images of beef cattle and their associated segmentation masks, the generative model can generate new images with realistic segmentation. These synthetic images are used to augment the training data for the facial recognition model, which improves the ability to accurately identify regions of interest (ROIs).

[0070] In one embodiment, the generative model is used to improve the facial recognition model. For example, the generative model can be used to generate diverse and realistic facial images of beef cattle, capturing variations in pose, lighting, and background. These synthetic images can be used to augment the training data set for the facial recognition model, thereby improving the model's ability to accurately identify individual beef cattle. By providing a more diverse and representative training data set, the facial recognition model achieves better generalization and performance in real-world scenarios.

[0071] Therefore, generative models (e.g., GANs, VAEs, etc.) provide a data augmentation method for developing facial recognition solutions for animals. By generating synthetic images with, for example, bounding box annotations, segmentation masks, and diverse beef cattle faces, the generative models improve the overall performance of the system 10. Integrating data augmentation into the data pipeline significantly enhances the generalization ability of the model, which results in more accurate and robust recognition of animals under a variety of conditions.

[0072] In some embodiments, an improved generative model is utilized that provides higher quality synthetic data that better simulates real-world changes. In some embodiments, domain adaptation is utilized. The use of domain adaptation techniques helps to transform knowledge from synthetic data into real-world data, thereby further improving the performance of system 10. In some embodiments, multimodal data augmentation is utilized. For example, multimodal data augmentation (such as combining visual and depth information) using a generative model can enhance the ability of system 10 to handle complex scenes. In some embodiments, evaluation metrics and methodologies are developed to quantify the impact of data augmentation based on a generative model on the performance of system 10. Using a generative model for data augmentation in an animal face recognition system provides a more effective and efficient recognition system.

[0073] 4. Models and comparisons

[0074] In some embodiments, system 10 utilizes one or more machine learning models. For example, in one embodiment, a first model is trained to detect faces and / or tags of animals (e.g., Figure 2 A cow face and label detection model 50 is used, and a second model is trained to calculate unique vector data for animal faces (e.g., Figure 2 In some embodiments, at least one machine learning model converts an image of an animal into a data vector, which can then be saved in a database or used to compare with other data vectors to assess similarity. In some embodiments, the comparison of the data vectors includes determining a distance score between two data vectors.

[0075] In one example, computer 26 executes stored instructions to compare the converted digital image to the stored digital image using a concurrent vector comparison. In some embodiments, the concurrent vector comparison calculates a degree of similarity between the converted digital image and the stored digital image.

[0076] In some embodiments, concurrent vector comparison uses a vector of values ​​to simultaneously compare multiple values ​​in different vectors, which can consist of a triplet loss technique in which a reference input image is compared with a matching input image, and a non-matching input image to minimize the difference between the reference input image and the matching input image and maximize the distance between the reference input image and the non-matching input image.

[0077] In some embodiments, the concurrent vector comparison utilizes a comparison algorithm to calculate differences between different animal faces, and using the difference calculation may select a stored image that is similar to at least one transmitted image or may indicate that no similar image exists in the non-transitory memory.

[0078] In one example, computer 26 executes stored instructions to send the calculated similarity to image capture device 14 using communication channel 22 .

[0079] In some embodiments, feedback (e.g., user feedback) is utilized to improve the accuracy of the model and improve the performance of system 10. In one embodiment, system 10 receives feedback regarding the accuracy of identifying a transmitted image as similar to a stored image. In one example, computer 26 executes stored instructions to receive feedback regarding the accuracy of a calculated similarity between at least one converted digital image and at least one stored digital image.

[0080] In this case, the comparison algorithm will be updated to improve the calculation, adjust the transformation, or, for example, adjust the randomness of the stochastic model or the margin processing quantified by the slack variables used in the triple method. Generally, in some embodiments, the improvement of the image comparison is improved by feedback, thereby improving the accuracy of recognizing animal faces. As an example, the computer 26 executes stored instructions to update the concurrent vector comparison calculation to improve accuracy.

[0081] 5. Animal Registration

[0082] In one embodiment, reference Figure 2 , a user inputs (e.g., registers) a known animal face into the system 10. After being registered, the known animal face can be used later to identify an unidentified animal. Stage 1 uses a detection model 50 to process an image of the animal. Stage 2 uses an embedding model 54 to process the detected face, which represents the image as vectorized data. In other words, the input known animal is processed and / or transformed to create a data vector that is stored for subsequent comparison. In other words, the vectorized data of the image of the known animal is then used for comparison.

[0083] refer to Figure 7 For example, the animal registration phase uses a camera 58 to capture images of animals (e.g., beef cattle) in a chute or grazing in a pasture or feedlot. In some embodiments, the camera 58 is a smart device with edge processing that is configured for real-time animal detection, cropping of animal heads, key point detection, and image upload. In some embodiments, the camera 58 is integrated with an animal holding mechanism (such as a chute). In some embodiments, the camera 58 is movable relative to the animal holding mechanism so that the camera 58 can capture images of the animal from different angles. The animal registration phase can also use a user device 62 that is capable of capturing images of the animal. The camera 58 and / or the user device 62 provides input data to a model 66 (e.g., a neural network), which provides a unique embedding 70 of the vectorized data to represent a known animal.

[0084] 6. Animal Identification

[0085] In one embodiment, reference Figure 3 , images of unidentified animals 74 entering the system are processed by a first model (“Stage 1”) trained to detect the face and / or label of the animal (e.g., Figure 2 A cow face and label detection model 50 is used, and a second model is trained to calculate unique vector data for animal faces (e.g., Figure 2 The data vector is then used to compare animal faces in images to identify positive and negative matches.

[0086] In one embodiment, reference Figure 8 , the image 78 of the unidentified animal is cropped, augmented, and key points are identified. In the illustrated embodiment, augmentation includes, for example, rotation and brightness adjustment. The key points and augmented data are used as input to a neural network model 82, which determines a unique animal embedding of the vectorized data 86 of the unidentified animal in the image.

[0087] refer to Figure 6 , the images are compared so that the distance or difference between the images is larger for negative matches and smaller for positive matches. In some embodiments, such distance or difference may be within an error range.

[0088] The results of the recognition are used to improve, for example, the embedding model. In some embodiments, the system uses the results to further improve the recognition by adjusting the distance or difference tolerance in the comparison of the vector data, where these distances can be individual differences or the sum of differences in the compared data. Figure 5 , making the vectorized data of positive matches more similar, and making the vectorized data of negative matches more different.

[0089] 7. Methods and Systems

[0090] As disclosed herein, system 10 is configured to collect information about known animals and identify unknown animals.

[0091] refer to Fig. 9 , the animal identification method 90 includes an animal registration phase 94 and an animal identification phase 98. In some embodiments, the registration phase 94 is completed before the animal identification phase 98. In other embodiments, the animal identification phase 98 is performed after some but not all of the animal registration phase 94 is completed.

[0092] In one embodiment, the animal registration stage 94 includes: (step 101) capturing at least one image of a first known animal with a device. In some embodiments, capturing at least one image of the first known animal includes: capturing a first image of the first known animal at a first perspective and capturing a second image of the first known animal at a second perspective (e.g., multiple images and different perspectives). In some embodiments, capturing at least one image of the first known animal includes capturing a video of the first known animal.

[0093] The animal enrollment phase 94 also includes (step 102) determining a first data vector based on at least one image of a first known animal using a model. In some embodiments, the model is a machine learning model. In some embodiments, the model is trained with an augmented data set (e.g., generated data, synthetic data, etc.). In some embodiments, the augmented data set includes rotated images, scaled images, flipped images, brightness adjusted images, generated images, or any combination thereof.

[0094] The animal registration phase 94 also includes (step 103) receiving a first set of identification information of a first known animal on the device. In some embodiments, the set of identification information includes an ear tag, a batch ID, a sex, a note, feeding performance, lameness, disease, antibiotic status, pen movement, or any combination thereof. In some embodiments, the set of identification information is entered by a user on the same device used to capture the image of the animal.

[0095] The animal registration phase 94 also includes (step 104) saving the first data vector and the first identification information set of the first known animal to the database. The animal registration phase 94 continues and is repeated for other known animals at (step 105). For example, the animal registration phase 94 also includes: using the device to capture at least one image of a second known animal, using the model to determine a second data vector based on the at least one image of the second known animal, receiving a second identification information set of the second known animal on the device, and saving the second data vector and the second identification information set of the second known animal to the database.

[0096] In one embodiment, the animal identification phase 98 includes: (step 106) capturing at least one image of the unidentified animal with the device; (step 107) determining a new data vector based on the at least one image of the unidentified animal using the model; (step 108) comparing the new data vector with the first data vector and the second data vector; and (step 109) identifying the unidentified animal as one of the animals registered in the animal registration phase 94 (e.g., the first known animal, the second known animal, etc.). In some embodiments, the animal identification phase 98 also includes (step 110) displaying the identified animal including the corresponding identification information set on the device. In some embodiments, the animal identification phase 98 also includes receiving feedback on the device regarding the accuracy of identifying the unidentified animal as the identified animal. The animal identification phase 98 continues and is repeated for other unidentified animals at (step 111).

[0097] Various changes and modifications to the disclosed embodiments will become apparent to those skilled in the art.

Claims

1. An animal identification method, comprising: Animal registration stage, the animal registration stage includes: capturing at least one image of a first known animal using a device, determining a first data vector based on at least one image of the first known animal using a model, receiving, on the device, a first set of identification information of the first known animal, storing the first data vector and the first identification information group of the first known animal in a database, capturing at least one image of a second known animal using the device, determining a second data vector based on at least one image of the second known animal using the model, receiving, at the device, a second set of identification information of the second known animal, and storing the second data vector and the second identification information set of the second known animal in the database; and The animal identification stage includes: capturing at least one image of an unidentified animal using the device, determining a new data vector based on at least one image of the unidentified animal using the model, comparing the new data vector to the first data vector and the second data vector, and identifying the unidentified animal as the first known animal or the second known animal, and The identified animal including the corresponding set of identification information is displayed on the device.

2. The method of claim 1 , wherein capturing at least one image of the first known animal comprises: A first image of the first known animal is captured at a first viewing angle and a second image of the first known animal is captured at a second viewing angle.

3. The method of claim 1 , wherein capturing at least one image of the first known animal comprises: Capturing video of the first known animal.

4. The method of claim 1, wherein the animal identification stage further comprises: Feedback is received at the device regarding the accuracy of identifying the unidentified animal as the identified animal.

5. The method of claim 1, wherein the animal is a cow, sheep, horse, pig, goat, chicken, dog or cat.

6. The method of claim 1, wherein the first identification information group comprises: Ear tag, batch ID, sex, notes, feeding performance, lameness, disease, antibiotic status, pen moves or any combination thereof.

7. The method of claim 1, wherein the model is a machine learning model.

8. The method of claim 8, wherein the model is trained using an augmented dataset.

9. The method of claim 8, wherein the augmented data set comprises: Rotate an image, scale an image, flip an image, brightness adjust an image, generate an image, or any combination thereof.

10. An individual animal identification system comprising: an image capture device capable of capturing at least one digital image; a communication channel connected to the image capture device; as well as a computer connected to the communication channel, the computer comprising at least one processor and a non-transitory computer-readable memory; wherein the image capture device captures the at least one digital image and transmits the at least one digital image to the computer using the communication channel, thereby causing the computer to: detecting the presence of an animal face in the at least one digital image sent from the image capture device; determining and storing coordinates of a location of a face of an animal detected in the at least one digital image; applying at least one transformation of the detected animal face in the at least one digital image to create at least one transformed digital image; comparing the at least one converted digital image to at least one stored digital image using a concurrent vector comparison to calculate a similarity between the at least one converted digital image and the at least one stored digital image; sending the calculated similarity to the image capture device using the communication channel; receiving feedback regarding the accuracy of the calculated similarity between the at least one converted digital image and the at least one stored digital image; as well as The concurrent vector comparison calculation is updated to improve accuracy.

11. The system of claim 10, wherein the image capture device captures data including a three-dimensional orientation of objects in the digital image.

12. The system of claim 10, wherein the image capture device captures data including depth of objects in the digital image.

13. The system of claim 10, wherein detecting the presence of an animal face in the at least one digital image sent from the image capture device comprises: Key points in the image are detected, including eyes, nose, ears, mouth, and other known facial features of the animal.

14. The system of claim 10, wherein detecting the presence of an animal face in the at least one digital image sent from the image capture device comprises: Detect the depth of the animal.

15. The system of claim 10, wherein the concurrent vector comparison uses a vector of values ​​to simultaneously compare multiple values ​​in different vectors, which can consist of a triplet loss technique in which a reference input image is compared with a matching input image, and a non-matching input image to minimize the difference between the reference input image and the matching input image and maximize the distance between the reference input image and the non-matching input image.

16. The system of claim 10, wherein the at least one transformation of the detected animal face in the digital image comprises at least one of: cropping to eliminate non-facial data, orienting the detected animal face for standard analysis and comparison, scaling an image of the detected animal face to a standard size, detecting standard facial features, such as the distance between eyes, and deriving at least one vector of image data from the detected animal face.

17. The system of claim 10, wherein the at least one digital image can be a series of digital images including a video image.

18. The system of claim 10, wherein the concurrent vector comparison utilizes a comparison algorithm to calculate differences between different animal faces, and using the difference calculation can select a stored image that is similar to the at least one transmitted image or can indicate that a similar image does not exist in the non-transitory memory.