The Newest Trends in AI Pet Tech宠物AI技术的最新趋势

In recent years， technological advances have touched nearly every human industry. Now the pet industry is starting to benefit， with artificial intelligence （AI） at the heart of new technologies designed to boost the safety and quality of life for companion animals worldwide.

近年来，科技进步几乎触及各行各业。如今，宠物行业也开始获益，一系列以人工智能（AI）为核心的新技术问世，这些技术旨在提高世界各地宠物的生命安全与生活质量。

The enabling algorithm

使能算法

AI’s implementation can be covert （the matching algorithm finding suitable dog sitters in the area， say） or overt—like cat flaps that use computer vision to identify the animal trying to enter the house. The more ambitious implementations could genuinely transform the industry， as well as the lives of pets entrusted to algorithmic assistance.

AI的应用可以是间接的（譬如，用匹配算法寻找附近合适的狗保姆），也可以是直接的——在计算机视觉技术的加持下，活动板猫洞可以识别试图进入房子的动物。那些更大胆的应用则可能真正改变宠物行业，在算法辅助的照料下，宠物的生活也将焕然一新。

One of the most exciting frontiers in pet AI is the world of noseprint recognition. In South Korea alone， there are 2 firms working towards this： Petnow1， an innovation award winner at the 2022 CES tech trade show organized by the Consumer Technology Association， and iSciLab2， a company that made its name with its highly accurate human iris recognition.

宠物AI最令人兴奋的前沿领域之一是鼻纹识别技术。仅在韩国，就有两家公司致力于此：2022年美国消费者技术协会举办的技术贸易展创新奖得主派诺公司，和以高精度人类虹膜识别技术著称的iSciLab公司。

Non-invasive tracking

无创追踪

Like a human’s fingerprints， a dog’s noseprint is unique and permanently fixed from around the age of 6 months， making it a potentially transformational way of reuniting missing animals with their owners.

和人类的指纹一样，狗的鼻纹独一无二，在狗约6个月大时就会永久定型，因此，鼻纹识别有望成为帮助主人找回丢失宠物的一种变革性方式。

Obviously， microchips have already performed this function for decades， but as Peter Jung， Business Development Manager at Petnow explains， uptake3 isn’t mandated in every country. Noseprint recognition is an inexpensive and non-invasive way of getting the same positive outcome， using nothing more than the phone in your pocket.

当然，微型芯片用于寻找丢失宠物已经有几十年了，但派诺的业务拓展经理彼得·郑解释道，并非每个国家都强制要求宠物植入芯片。鼻纹识别便宜、无创，只需使用口袋中的手机便可实现，效果和微型芯片相同。

“Noseprints are just with them， and they cannot be damaged or lost like tags or microchips，” Jung says， noting that criminals sometimes remove microchips to resell stolen animals without being caught.

郑表示：“鼻纹就在狗狗身上，不会像狗牌或芯片那样损坏或丢失。”他指出，犯罪分子有时会先取出被盗宠物体内的微型芯片，再将宠物重新出售，就能不被抓获。

With anyone able to identify a registered animal using just a smartphone camera and the app， he sees this as a way of lowering pet insurance premiums， and moving towards the ultimate goal： a world without lost， abandoned or stolen pets.

只要使用智能手机摄像头和派诺的应用程序，就能识别注册过的宠物。郑认为，鼻纹识别技术能降低宠物保险费，并接近最终目标：一个宠物不再丢失、遗弃或被盗的世界。

Nose recognition—the pros and cons

鼻纹识别——利与弊

iSciLab’s ambition is similar， and the company plans to commercialize both in Korea and abroad. “Having attained 99.99% matching accuracy for dogs， we are also seeking to progress our technology to cover cats as well as cattle in the near future，” says the company’s Chief Operating Officer， Meenho Sheen.

iSciLab的目标和派诺类似，该公司计划在韩国内外实现鼻纹识别的商用。“我们对狗的鼻纹匹配准确率已达到99.99%。目前，我们正尝试将这一技术应用于猫和牛，希望在不久的将来能够实现。”iSciLab的首席运营官辛敏浩（音译）表示。

Petnow is also working on enabling cat nose recognition this year—a tougher proposition due to the smaller size and relative shallowness to a dog’s， explains Jung.

郑解释道，派诺公司今年也在努力实现猫的鼻纹识别。这是一项更为艰巨的任务，因为猫鼻比狗鼻更小、更浅。

Though exciting， there are 2 possible drawbacks to noseprint recognition replacing the traditional microchip. The first is that getting a dog to stand still enough to scan a noseprint isn’t always straightforward—especially if it’s a stray that doesn’t trust you. The second drawback is also of a practical nature： to be effective， a noseprint database needs a critical mass of people to register and use the app.

尽管这一切令人兴奋，但用鼻纹识别取代传统的微型芯片可能存在两个不足。首先，让一只狗——尤其是不信任你的流浪狗——站着不动扫描鼻纹，有时并不容易。 另一个不足也和可行性有关：有效的鼻纹数据库需要应用程序有大量用户注册和使用。

The medical approach

医疗路径

Other players are looking at the medical potential of AI. Dogiz， a British canine health-tracking app， has a feature called Dr. Poop that will analyze the pet’s stools for medical red flags4 that could require vet attention. Korean TTcare， meanwhile， uses this science to look for pet conditions visible via an ocular or skin analysis.

其他公司则关注AI的医疗潜力。英国犬类健康追踪应用程序Dogiz有一个叫“便便博士”的功能，可以对宠物的粪便进行分析，发现可能需要就医的危险信号。与此同时，韩国宠物健康应用程序TTcare则使用该技术寻找眼部或皮肤分析能发现的病症。

“The AI has been trained with over 2 million pet image data samples with more than 90% accuracy as professional veterinarians continue to label and verify the signs on photos，” explains a spokesperson for AI For Pet， the company behind TTcare.

TTcare背后的研发公司AI For Pet公司的发言人解释道：“经过200多万个宠物照片数据样本训练，加上专业兽医不断标记、核对照片中的迹象，该AI模型准确率已达到90%以上。”

The company’s data shows that just over 47% of the dog-eye photos analyzed by the app have shown signs of abnormalities， with skin issues coming in second （42%）.

该公司的数据显示，在应用程序所分析的狗眼照片中，47%都有异常迹象，其次是皮肤问题（异常率42%）。

While the AI’s training data is sourced from shelters， pet cafes and vets， many of the data samples come from the users themselves. That’s the beauty of AI： as it acquires more data via regular users， its effectiveness grows.