The Newest Trends in AI Pet Tech寵物AI技術的最新趨勢

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.

近年來，科技進步幾乎觸及各行各業。如今，寵物行業也開始獲益，一系列以人工智能（AI）為核心的新技術問世，這些技術旨在提高世界各地寵物的生命安全與生活質量。

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的應用可以是間接的（譬如，用匹配算法尋找附近合適的狗保姆），也可以是直接的——在計算機視覺技術的加持下，活動板貓洞可以識別試圖進入房子的動物。那些更大膽的應用則可能真正改變寵物行業，在算法輔助的照料下，寵物的生活也將煥然一新。

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.

寵物AI最令人興奮的前沿領域之一是鼻紋識別技術。僅在韓國，就有兩家公司致力于此：2022年美國消費者技術協會舉辦的技術貿易展創新獎得主派諾公司，和以高精度人類虹膜識別技術著稱的iSciLab公司。

Non-invasive tracking

無創追蹤

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.

和人類的指紋一樣，狗的鼻紋獨一無二，在狗約6個月大時就會永久定型，因此，鼻紋識別有望成為幫助主人找回丟失寵物的一種變革性方式。

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.

當然，微型芯片用于尋找丟失寵物已經有幾十年了，但派諾的業務拓展經理彼得·鄭解釋道，并非每個國家都強制要求寵物植入芯片。鼻紋識別便宜、無創，只需使用口袋中的手機便可實現，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.

鄭表示：“鼻紋就在狗狗身上，不會像狗牌或芯片那樣損壞或丟失。”他指出，犯罪分子有時會先取出被盜寵物體內的微型芯片，再將寵物重新出售，就能不被抓獲。

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.

只要使用智能手機攝像頭和派諾的應用程序，就能識別注冊過的寵物。鄭認為，鼻紋識別技術能降低寵物保險費，并接近最終目標：一個寵物不再丟失、遺棄或被盜的世界。

Nose recognition—the pros and cons

鼻紋識別——利與弊

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的目標和派諾類似，該公司計劃在韓國內外實現鼻紋識別的商用。“我們對狗的鼻紋匹配準確率已達到99.99%。目前，我們正嘗試將這一技術應用于貓和牛，希望在不久的將來能夠實現。”iSciLab的首席運營官辛敏浩（音譯）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.

鄭解釋道，派諾公司今年也在努力實現貓的鼻紋識別。這是一項更為艱巨的任務，因為貓鼻比狗鼻更小、更淺。

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.

盡管這一切令人興奮，但用鼻紋識別取代傳統的微型芯片可能存在兩個不足。首先，讓一只狗——尤其是不信任你的流浪狗——站著不動掃描鼻紋，有時并不容易。 另一個不足也和可行性有關：有效的鼻紋數據庫需要應用程序有大量用戶注冊和使用。

The medical approach

醫療路徑

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.

其他公司則關注AI的醫療潛力。英國犬類健康追蹤應用程序Dogiz有一個叫“便便博士”的功能，可以對寵物的糞便進行分析，發現可能需要就醫的危險信號。與此同時，韓國寵物健康應用程序TTcare則使用該技術尋找眼部或皮膚分析能發現的病癥。

“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背后的研發公司AI For Pet公司的發言人解釋道：“經過200多萬個寵物照片數據樣本訓練，加上專業獸醫不斷標記、核對照片中的跡象，該AI模型準確率已達到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%）.

該公司的數據顯示，在應用程序所分析的狗眼照片中，47%都有異常跡象，其次是皮膚問題（異常率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.

該AI模型的訓練數據源自動物收容所、寵物咖啡店、獸醫診所，但很大一部分數據樣本都來自用戶自身。這就是AI的魅力：從普通用戶處獲得的數據越多，使用效果就越好。

Indeed， initially， the app could only detect 4 irregularities in dogs’ eyes. Already that number has risen to 11， thanks to an ever-growing data set. TTcare plans to expand not only the number of abnormalities but also the areas that can be checked， such as joints and teeth.

事實上，這款應用程序起初只能發現4種狗眼異常情況，但由于數據集不斷擴大，現在已經能發現11種。TTcare不僅打算增加異常情況種類，還計劃增加可檢查部位，譬如關節和牙齒。

A niche5 segment with potential

具有潛力的商機

Within the consumer pet market， AI is still very much in its infancy， and as such， getting market projections on it is tricky—in part because AI can be anything from the unique selling point （USP） of a product， to a background feature underpinning something else.

在寵物消費市場，AI的應用在很大程序上才剛起步。嚴格地說，要做出準確的市場預測并不容易——部分原因在于，AI用途很廣，既可以是某產品的獨特賣點，也可以是支撐另一產品的背景功能。

According to Kim Bill， Head of Purina’s Accelerator Lab6， around a third of the hopeful start-ups vying for the company’s support use AI in some capacity. These include “a couple of applications with nose recognition，” she says.

普瑞納公司的“加速器實驗室”負責人金·比爾表示，在競相爭取普瑞納公司支持的眾多前景廣闊的初創企業中，約有三分之一或多或少會使用AI，包括“好幾個帶鼻紋識別功能的應用程序”。

Other examples include everything from computer vision programs to analyze pets’ body language， to Internet of things （IoT） devices tracking behavior. Indeed， Purina has its own take on7 the latter： a connected litter box that can keep an eye on cats’ behavior for early warnings of ailments from Urinary Tract Infection （UTI） to hyperthyroidism8.

還有很多其他的AI應用，包括分析寵物肢體語言的計算機視覺程序、追蹤寵物行為的物聯網設備等。關于后者，普瑞納公司有自己的想法：他們推出了一款物聯網貓砂箱，可監測貓咪行為，以發現尿路感染、甲狀腺功能亢進等疾病的早期征兆。

The limitations of technology

科技的局限

Generally upbeat about9 AI’s place in the pet market， Bill remains somewhat concerned about misdiagnoses made by medical tech. An AI-enabled IoT pet collar might report a dog with a dangerously high fever， when it really just sat near a radiator， for example. Worse， such tech could lull owners into10 a 1 sense of security if something is wrong that AI can’t yet detect.

AI能否在寵物市場占有一席之地？總體上，比爾持樂觀態度，但醫療技術的誤診問題仍讓她有些擔憂。譬如，AI物聯網寵物項圈可能會報告一只狗嚴重高燒，但事實上它只是坐在暖氣片旁邊。更糟糕的情況是，如果愛寵身體確實有問題，但現階段AI還無法發現，主人就有可能因此產生虛假的安全感。

But she still believes there’s a market for quality products that can help humans understand their animals’ needs. Just as long as AI doesn’t stand in the way of the essential emotional connection between pet and owner.

但比爾仍認為，幫助人類了解寵物需求的高質量產品是有市場的，只要AI不會妨礙寵物與主人之間的必要情感聯系。

“I think if it’s used to replace that bonding element between pet owner and pet， then you know， it probably would not have such a high uptake，” she concludes.

她總結道：“我覺得，如果AI被用來替代主人與寵物之間的感情紐帶，可能就不會有那么多人接受它了。”

（譯者為“《英語世界》杯”翻譯大賽獲獎者；單位：中國人民大學外國語學院）

1派諾公司，提供寵物服務提供商的資料，并幫助寵物主人預定寵物看護、遛狗、美容、訓練等服務。

2 iSciLab公司，主要業務為人工智能和機器學習。" 3 uptake（對現有東西的）利用，應用。

4 red flag 紅旗，危險的信號，預示可能發生的危險。

5 niche（產品的）商機，市場定位，向特定人群銷售某個產品的機會。

6普瑞納公司是美國寵物食品制造公司，旗下的“加速器實驗室”主要幫助發展寵物護理初創企業。" 7 one’s own take on某人對……的看法。" 8 hyperthyroidism甲狀腺功能亢進。" 9 be upbeat about對……持樂觀態度。" 10 lull sb into麻痹、誘使某人產生某種想法。