“It took us about 20 seconds to find the first in the area designated by the model.” wrote Jaffer in a blog message, documenting a field test. Starting with the Milton public center, where the model showed high confidence in costumes near the parking lot, the team systematically visited places with various forecasting levels.
In the Milton Country Park, in each region with the high confidence that they checked, contained a significant growth of Brecht. When they explored a living hot point, they found that an empty plot was full of bran. The most funny is that a major prediction in Northern Cambridge led them to Bramblefields Local Natural ReserveThe field is true to its name, the region contained an extensive coating of Bramble.
The model was reported best when the large, naked Bramble spots are found, visible from above. Smaller offers under the cover of trees showed lower trust indicators – a logical restriction, given the prospect of invoices of the satellite. “Since Tessera is studying the presentation from remote sensing data, it would make sense that Bramble is partially hidden from above,” Jaffer explained.
Early experiment
While researchers expressed enthusiasm for early results, Bramble detection is evidence of a concept that is still in an active study. The model has not yet been published in a reviewed journal, and the fields described here was an informal test, and not a scientific research. The Cambridge team recognizes these restrictions and plans for a more systematic verification.
Nevertheless, this is still a relatively positive study of the application of the methods of the neural network, which reminds us that the field of artificial intelligence is much larger than just generative models of AI, such as ChatGPT or models of video synthesis.
If the study of the team disappears, the simplicity of the Bramble detector offers some practical advantages. Unlike more intense resource models of deep training, the system can potentially work on mobile devices, which provides a field test in real time. The team considered the question of developing an active telephone telephone system based on a phone, which will allow researchers of field flights to improve the model, while checking its forecasts.
In the future, similar approaches based on artificial intelligence, combining satellite remote sensing with civil science data, can potentially display invasive types, track agricultural pests or track changes in various ecosystems. For threatened species, such as hedgehogs, quickly mapping the critical characteristics of the habitat becomes more and more valuable at a time when climate change and urbanization actively change the places that a hedgehog like to call the house.
