A new facial recognition program can help out where human eyes might fail.
To many of us, let’s admit it, chimps look pretty much alike in their physiognomy. Then again, they themselves can easily tell one another apart… and so can new facial recognition software.
Developed by two American researchers and appropriately called ChimpFace, the facial recognition software has been designed to recognize new photos of chimpanzees published on social media and e-commerce websites. Better yet: it can also recognize the individual chimps in them.
The aim is to identify and track chimps that are being trafficked online. Once suspicions of trafficking are aroused, posts will be flagged and sent to conservation experts for review.
“The initial phase of the project will focus on chimpanzees because of their importance to conservation, the frequency at which they appear for sale online, and the availability of large amounts of imagery,” explain the software’s developers, conservationists Alexandra Russo and Dr. Colin McCormick on the project’s website.
“We will leverage both publicly available imagery dataset and images we have collected from chimpanzee conservation and research organizations to build a binary chimpanzee image classifier, and then deploy it to monitor social media posts,” they go on. “Depending on the number of posts that are flagged by the image analysis method, we may also add text-analysis to our method to down-select posts based on likely trafficking-related words or phrases (i.e. ‘chimp for sale’).”
Wildlife trafficking is a multi-billion dollar business and traffickers worldwide have been resorting increasingly to the Internet to sell endangered “exotic” animals. Juvenile animals that are widely seen as cute, including baby chimps, are frequently sold online as part of the exotic pet trade.
High-tech tools can help roll back the online trade. “However, monitoring is expensive and difficult to do manually,” the two experts point out. “Live great apes frequently fall victim to Internet-facilitated trafficking. Over 6,000 great apes have been trafficked from the wild since 2005, and a further 30,000 have died in capture-related activities,” they add.
“Researchers have identified hundreds of social media accounts displaying illegally traded great apes, and have manually searched millions of social media photographs for evidence of trafficking,” they elucidate. “However, this method is cumbersome and prone to error, as many individual great apes are similar in appearance, making them difficult to track through the online trade chain as well as susceptible to misidentification and/or double counting.”
This new facial recognition software will help out where human eyes often fail.