- The endangered humphead wrasse, a reef fish that swims the seas from Africa to the South Pacific, is in high demand in mainland China and Hong Kong as a luxury culinary delicacy.
- Despite harvest limits, trading regulations and fishing bans, it’s overfished and illegally traded.
- Researchers in Hong Kong have developed a new AI-based photo identification smartphone app, Saving Face, to help enforcement officers identify individual fish using their unique facial patterns with just a photo.
- Researchers say they hope the app can address both illegal laundering of humphead wrasse and mislabeling of wild-caught fish as captive-bred; its developers say it can be tweaked to identify other species that have unique markings.
Among the many spectacular reef fishes swimming the Indo-Pacific corals, from Africa to Oceania, is the big-eyed, bulge-faced, thick-lipped, blue-green humphead wrasse, also known as the Napoleon wrasse. It’s among the world’s largest reef fish, sought for its flavor and coveted as a status symbol meal in high-end Chinese cuisine.
Overfishing remains the biggest threat to the species, Cheilinus undulatus, whose conservation status was assessed as vulnerable in 1996 on the IUCN Red List, and worsened to endangered in 2004. Juvenile and young adult humphead wrasses are sold alive to restaurants across Hong Kong and mainland China, as well as to expatriate Chinese communities around the world. Customers choose the fish for their meal from the wrasse swimming in large tanks, much like U.S. restaurant-goers choose a lobster for dinner.
The humphead wrasse is extremely valuable: each fish can fetch up to $850. While there is a legal trade in the species, the high market value has also led to a booming illegal trade. Permits that allow shops and restaurants to keep legally caught wrasse are often misused to launder illegally traded fish, as there has been no way to identify individuals. Indonesia is the leading exporter of humphead wrasse.
In a recent study, published in the journal Frontiers in Ecology and Evolution, researchers in Hong Kong announced a first-of-its-kind photo identification app for the humphead wrasse, called Saving Face. They developed this smartphone-based app, which uses artificial intelligence (AI), to identify individual humphead wrasses from photos. It pulls the most likely matches from a repository of images based on the fish’s unique facial markings: distinctive black lines around their eyes and squiggly patterns on their faces.
Each wrasse’s markings are asymmetrical, with patterns that differ on each side of their face and remain unchanged for years, which makes it easy to identify individuals. That’s why the app is a great tool for identification, said wildlife trade researcher and lead author Loby Hau. The markings function “just like people’s fingerprints … very complex and also unique,” Hau, who worked on the app as part of his doctorate at the University of Hong Kong’s Swire Institute of Marine Science, told Mongabay.
Scientists have long used a variety of approaches to identify individual animals of different species. “But until this point, relatively few have been used practically within the context of wildlife trade,” Alice Hughes, a University of Hong Kong wildlife trade researcher, told Mongabay in an email. This app, she said, “facilitates the tracking and tracing of wildlife, and enables enforcement and the identification of illegal trade.” Hughes wasn’t involved in the Saving Face research.
Tracking fish to address loopholes
The humphead wrasse is the first reef fish to be listed on Appendix II of CITES, the international convention on the wildlife trade, which allows trade that’s regulated in a way that’s not detrimental to wild populations. The legal commercial trade is regulated through export/reexport permits.
Hong Kong requires additional licensing for retail outlets that keep live CITES Appendix II species. Hong Kong’s Agriculture, Fisheries and Conservation Department (AFCD), which oversees the trade, stipulates the maximum number of animals that sellers can keep at any given time during a five-year license period. In the case of humphead wrasses, the outlets must keep trade records of each fish for three years so that officials can trace back and verify exports, sales and reexports in case of an investigation, since these animals are ultimately consumed.
However, there are loopholes in the system. Without tags or other identification, Hau said, it’s not possible to track individual humphead wrasses coming into or being traded out of a shop or restaurant. Traders can readily replace a legally obtained wrasse, with trafficked fish while staying within the license quota. Sometimes wild-caught humphead wrasses are mislabeled as ranched: a system where young, legally imported wild fish are grown in captivity to better adjust to being in a restaurant tank.
The Saving Face app, developed in partnership with the AFCD, can identify individual humphead wrasses in the trade, plugging this loophole. It’s precise: with a good-quality photo, it can identify the exact match in the top three results 95% of the time. With images of both sides of a fish’s face, the accuracy jumps to 99%. Now, Hong Kong officials are trying out a pilot version of the app, Hau said.
It’s an interesting application of technology, said humphead wrasse expert Rick Hamilton at The Nature Conservancy, who wasn’t involved in the app’s development. The basic premise here, he said, “is similar to when you go through customs and they take a photo of your face and recognize you against your passport … which is, I think, the first time that’s been done in [regulating] the wildlife trade.”
This technology could potentially be used widely. The Saving Face developers say their AI-based photo identification model can be modified to recognize a range of species if they have unique visible markings.
With the $20 billion illegal wildlife trade hastening the demise of innumerable species across the globe, experts say new tools are crucial to curbing it. Hamilton said apps like Saving Face are the future. “It’s hard to see how tools like this won’t become more prevalent and more applicable as the technology advances.”
Hughes said the international trade in marine species, such as tropical fish, is frequently overlooked, and monitoring the trade results in better conservation. “Tighter regulation of live reef fish overall would likely make it harder to launder fish,” she said.
In addition to helping enforcement officers, Saving Face also offers a public-facing version intended for citizen scientists and fish enthusiasts who want to learn more about marine conservation.
The app’s ability to recognize individuals, Hau said, drives the idea that fish are wildlife, not just food. “Sometimes people may think endangered species may be … very rare or expensive, like ivory or rhino horn, but actually you can also see them in your daily lives.”
Banner image: The humphead wrasse is one of the largest reef fish that can grow to about 1.5 metres (about 5 feet) and weigh up to 190 kilograms. But juveniles and young adults are preferred for the trade. Image by Diego Delso via Wikimedia Commons (CC BY-SA 4.0).
AI and satellite data map true scale of untracked fishing and ocean industry
Citations:
Hau, C. Y., Ngan, W. K., & Sadovy de Mitcheson, Y. (2025). Leveraging artificial intelligence for photo identification to aid CITES enforcement in combating illegal trade of the endangered humphead wrasse (Cheilinus undulatus). Frontiers in Ecology and Evolution, 13. doi:10.3389/fevo.2025.1526661
Hau., C. Y., & Sadovy de Mitcheson, Y. J. (2023). Mortality and management matter: Case study on use and misuse of ‘ranching’ for a CITES Appendix II-listed fish, humphead wrasse (Cheilinus undulatus). Marine Policy, 149, 105515. doi:10.1016/j.marpol.2023.105515
FEEDBACK: Use this form to send a message to the author of this post. If you want to post a public comment, you can do that at the bottom of the page.
Credits
This story first appeared on Mongabay
This article is licensed under a Creative Commons Attribution-NoDerivatives 4.0 International License.
You may republish this article, so long as you credit the authors and Mongabay, and do not change the text. Please include a link back to the original article.
