Home > Article > Technology peripherals > Has facial recognition for pets been implemented?
Translator|Cui Hao
Reviewer|Sun Shujuan
Technology that can accurately identify animals will help owners reunite with lost pets, farmers monitor their livestock, researchers study wild animals. Based on the above application scenarios, microchips are the most popular pet identification method. However, implanting a chip requires invasive surgery, specialized equipment to read the chip, and the risk of a thief extracting the microchip. Another method is DNA analysis, which although accurate, is also very expensive and time-consuming. Today I want to introduce to you how to identify animals through facial recognition.
Face recognition of pets by computer vision solutions can be used as an alternative to the above solutions. Despite its shortcomings, this technology can demonstrate a high degree of accuracy in certain situations.
So, how does pet face recognition work? What are the challenges hindering technological progress? How does pet facial recognition work?
Generally speaking, pet facial recognition solutions are mainly divided into three steps:
(1) Image capture: Photos of animals are taken by high-resolution cameras. Some algorithms only work with predefined poses, so it is necessary to select images that meet these criteria.
(2) Feature extraction: Evaluate the suitability of the animal’s biometric data and perform preprocessing if necessary. The algorithm then extracts the feature set required for recognition.
(3) Feature matching: The extracted features are mathematically represented and matched with other images. For example, if we were looking for a dog in a lost pet database, we would match the dog's unique characteristics to all the animals in that database.
There are several ways to perform matching. One method is to use algorithms such as KNN and DBSCAN for clustering, which will produce a set of images that are highly close to our target image, and the user can manually select the most suitable image.
Additionally, probabilistic methods can be deployed to express the final results as confidence levels. For example, if the recognition target is a cat, and the algorithm determines that the captured image is a cat, its confidence level is 90%; if it is determined to be a dog, its confidence level is 10%.
Pet facial recognition
Here are examples of how to use pet facial recognition in real life. Some of these systems are successful, popular, or even commercial solutions, and some are academic attempts to test hypotheses.
Finding Lost Animals
Losing their pets is heartbreaking for owners. According to statistics, such incidents are more common than people think. In the United States, 80% of all pets (dogs/cats) owned by one-third of households are never found. There are several pet facial recognition programs that can help owners find their lost friends.
ForPaws
ForPaws is an APP product that uses pet identification based on the tip of the dog’s nose and skin color. and fur type to identify them. Owners must upload at least three photos to create a pet's profile. Currently, the solution can identify 130 dog breeds with an accuracy of 90%.
PIP
This pet identification company has developed an app that allows pet owners to register and upload photos of their animals. The system analyzes your pet's facial features. PiP claims it will be able to identify every lost cat and dog if the owner can provide additional information, such as gender, size and weight.
Anyone who finds a lost pet can use the app to find the pet's owner. PiP’s solution also continuously scans social media for animal posts and sends lost pet alerts to residents in relevant communities.
Love Lost
Love Lost by Petco is an app for pet owners and animal shelters. Users can create animal profiles so that when a pet goes missing, the software can begin matching the pet's biometric information to new arrivals to the shelter and other candidate animals.
Identify a specific pet
The training algorithm can be used to identify a specific pet. For example, pet owners can accurately identify their pet and take subsequent actions, such as sending an alarm or opening a door to let their pet in.
Aakaitz Garro front-end engineer develops pet facial solution
Arkaitz Garro, a front-end engineer at WeTransfer, developed a pet facial recognition solution that can identify a neighbor’s cat and send an alert to Garro (or a designated person) when the little guy shows up at the door. To capture photos of cats, Garro used a small camera and a Raspberry Pi with motion detection software. When an animal approaches the camera, a photo is taken and sent to the AWS Recognition platform, which compares it to other images of the cat uploaded by Garro. If the match is successful, the engineer is notified.
Microsoft’s IoT device for pet identification
Similarly, Microsoft has developed an IoT device that users can install on Where pets enter the home. When a pet is recognized, the device will unlock to let the animal in.
Assisting Scientific Research - Facial Recognition of Dolphins
In addition to identifying household pets, facial recognition algorithms can be used to detect other species. A study published in the Journal of Marine Mammal Science investigated the set of characteristics needed to identify dolphins. Researchers tracked and photographed 150 bottlenose dolphins over 12 years. The team identified the dolphins' faces and dorsal fins throughout their life cycle and assessed the feasibility of this approach.
Using these 150 subjects, it came down to only 31 dolphins with comprehensive profiles (i.e. clear photos of the left and right sides of the face and dorsal fin). The study relied on human expert opinion and statistical methods to detect similarities between different images of the same dolphin.
The results of this experiment show that dolphin facial features remain consistent over time and can be used for identification purposes. This technology can identify adult dolphins and facilitate tracking of dolphin growth, which has greatly promoted dolphin research.
Characteristics of the same dolphin at different stages
Help farmers monitor livestock
Identifying farm animals can be challenging. Zhao Jinshi, founder of Beijing Yourui Technology, has been engaged in software development in the agricultural field for many years. He has his own views on identifying farm animals: "For pigs, it will be more difficult to identify. Because pigs all look the same, but cows will It's relatively easy. Because they are obviously special. The patterns on their bodies are black and white and have different shapes."
Cows and their "interaction"
However, a challenge arises when it comes to cow identification – namely determining where to install the cameras. Cows are curious animals, and even the slightest change in their environment will attract their attention. When they find a camera, they "interact" with it by licking it. But challenges aside, having a system in place that can identify individual cows would greatly help breeders. The solution matches an animal's health, drinking and eating patterns with the animal's identity. With the help of artificial intelligence, it will be possible to detect signs of disease and abnormal behavior in animals and notify breeders in emergencies.
If you want There are three main challenges to consider when implementing a pet facial recognition solution:
#Determining the optimal feature set
Scientists have specified the Feature vectors for face recognition. However, the same approach doesn't work for pets because we don't know which features need to be used and how to interpret them. For example, when working with people, scientists can use variational autoencoders (VAE) architectures to extract features from faces. In this method, photos of people are compressed into vectors containing desired features, such as skin tone and facial expressions. But in terms of pet facial recognition, there is currently no reliable feature vector.
Reliable Feature Vectors
Solving for reliable feature vectors will significantly advance the field. The open source example DogFaceNet is an implementation of dog recognition based on deep learning. It uses the dog's eyes and nose as a feature set. This solution works reasonably well if the overall goal is to differentiate between dog breeds, but its performance is rather poor when it comes to differentiating between twins.
Animal Pose
Another example is using the Local Binary Pattern Histogram (LBPH) algorithm, which converts images into pixels , and operates by comparing the pixel values of different images. This method depends on the animal's posture, which makes it sensitive to changes in posture. It's worth mentioning that this approach is outdated. Currently, engineers tend to use more advanced features.
Getting animals to pose for the camera
For humans, it's easy to assume a specific pose and sit still. However, it’s not so easy when we try to get a cat or dog to take a photo in a specific pose. This requires the application of "pose-sensitive" facial recognition algorithms, such as those that rely on pixel similarity.
Provide a comprehensive training data set
For training to be effective, the data must be diverse and cover all the tasks the algorithm is expected to perform. For example, if the algorithm recognizes different dog breeds, then the data set should contain sources of information captured from different angles and correctly labeled. A variety of things can happen here, for example: someone might submit an image of mixed breeds, someone might incorrectly tag the image and assign the wrong breed name. To avoid such problems, experts must review all photos in the dataset one by one to verify the legitimacy of the images and the accuracy of the labels.
Pet face recognition field Progress has been hampered by the fact that researchers still can't pinpoint the best combination of features that can be used to accurately identify animals on a large scale. Nonetheless, there are some successful applications working on restricted data, such as identifying a specific animal or a small group of domestic/wild animals. If you are building your own pet facial recognition system, remember that animals are non-cooperative users of biometric technology. Some will insist on licking the camera, others will refuse to pose for photos. To eliminate unnecessary trouble, posture- and expression-neutral algorithms can be designed. Another issue to consider is privacy regulations. If you want to build an app for finding lost pets, do you need the owner to reveal their location? Even photos of pets in an owner's home may reveal private information about the owner.
Original link: https://readwrite.com/pet-face-recognition-are-we-there-yet/
Cui Hao, 51CTO community editor and senior architect, has 18 years of software development and architecture experience and 10 years of distributed architecture experience. Formerly a technical expert at HP. He is willing to share and has written many popular technical articles with more than 600,000 reads. Author of "Principles and Practice of Distributed Architecture".
The above is the detailed content of Has facial recognition for pets been implemented?. For more information, please follow other related articles on the PHP Chinese website!