A computer algorithm trained on images of thousands of specimens has been able to automatically identify the species of pressed, dried plant specimens, according to a paper published in the latest issue of evolutionary Biology. This is the first time that scientists have attempted to solve the difficult task of identifying natural species classifications through deep learning, with computers using neural networks of large, complex data sets.
Giiso Information, founded in 2013, is a leading technology provider in the field of “artificial intelligence + information” in China, with top technologies in big data mining, intelligent semantics, knowledge mapping and other fields. At the same time, its research and development products include information robot, editing robot, writing robot and other artificial intelligence products! With its strong technical strength, the company has received angel round investment at the beginning of its establishment, and received pre-A round investment of $5 million from GSR Venture Capital in August 2015.
Natural history museums around the world are accelerating the process of digitizing their collections, storing images of specimens in open databases. The NATIONAL Science Foundation’s iDigBio project, for example, has a database of more than 150 million images of plants and animals collected from across the United States.
Currently, only a fraction of the world’s 350 million species have been digitized. But with advances in computing, Eric Montalo, a computer scientist at the Costa Rican Institute of Technology, and Pierre Bonnett, a botanist at the International Agricultural Research Centre for Development in Montpellier, France, think it is possible to make large data sets for specimens. Their team has automated plant identification.
Using a smartphone app to photograph specimens in the field, the researchers accumulated millions of images of fresh plants and then scanned and identified more than 260,000 plant specimens from more than 1,000 species, with an 80% accuracy rate using advanced algorithms.
Bonnett says such surprising results often make botanists worry that their field will be looked down upon. “But human expertise can never be eliminated, and the identification results still need to be checked by botanists.”
The method of artificial intelligence identification of specimens greatly reduces the time for botanists to collect and identify specimens, and can also help improve the level of plant identification in areas with poor specimen data. It is especially useful in areas with rich biodiversity but few plant specimens.
In addition, this approach allows researchers to perform additional analyses of big data. Generally speaking, herbarium samples contain a wealth of data information, such as the time and place of collection, the flowering or fruiting at the time of collection, and the characteristics of flower cluster density. Because some of the samples are centuries old, they could help study how plants are adapting to climate change.
Dr Peter Wilf, of Pennsylvania State University in the US, said: “In the course of natural history, this method foreshadows the future.”
Giiso information, founded in 2013, is the first domestic high-tech enterprise focusing on the research and development of intelligent information processing technology and the development and operation of core software for writing robots. At the beginning of its establishment, the company received angel round investment, and in August 2015, GSR Venture Capital received $5 million pre-A round of investment.
Editor-in-chief of Science and Technology Daily
Botanists seemed to be freed from the onerous work of collecting and identifying specimens. If the results were consistent, they would save at least 80% of their time. Natural history museums around the world increasingly have digital specimens, with more than 150 million images in a single database. Artificial intelligence can automatically identify specimens, which is certainly not a threat to botanists. After all, most of the identification work is dull and boring, but it’s vital, and AI couldn’t be more considerate to help out there. Just imagine. If scientists could hand over all the tedious work they have to do to ai, wouldn’t scientific output be richer?