- PhD Level
- Posts
- š½ AI tracks cornāless data!?
š½ AI tracks cornāless data!?
Daily news that is actually intellectually stimulating.
Anthony Ao
August 15, 2025
In partnership with
Meet your new assistant (who happens to be AI).
Skej is your new scheduling assistant. Whether itās a coffee intro, a client check-in, or a last-minute reschedule, Skej is on it. Just CC Skej on your emails, and it takes care of everything:
Customize your assistant name, email, and personality
Easily manages time zones and locales
Works with Google, Outlook, Zoom, Slack, and Teams
Skej works 24/7 in over 100 languages
No apps to download or new tools to learn. You talk to Skej just like a real assistant, and Skej justā¦ works! Itās like having a super-organized co-worker with you all day.
Intact introns enable fast, scarless circRNA: āCIRCā even circularizes fullālength dystrophin
A Beijing-led team reports an approach that preātrains a knowledgeāguided convolutional neural network (KGCNN) on synthetic canopy spectra generated by a 3D radiative transfer model (LESS), then fineātunes it with a small 2021 field dataset to estimate true leaf area index (LAI) from drone multispectral imagery. In multiāyear tests on the same maize site, the model generalized to 2022ā2023 and outperformed 1D RTM (PROSAIL) and other ML baselines: for 2022, overall RĀ² rose by 0.27 and RMSE fell by 2.46 versus PROSAIL+TL; for 2023, RMSE dropped by 1.62.
Stageāwise results show, for 2022, RĀ² of 0.42/0.53/0.44 (jointing/trumpet/tasseling) with RMSE as low as 0.49; in 2023, RĀ² was 0.34/0.16/0.38 (RMSE 0.38ā0.92). The model tended to underestimate LAI at dense canopy during tasseling, likely from spectral saturation. Withināyear validation after fineātuning on 2021 achieved RĀ²āÆ=āÆ0.87 (RMSEāÆ=āÆ0.54). Overall, KGCNN+TL delivered higher accuracy and stability than LSTM, RNN, and RF alternatives while relying only on fiveāband UAV data.
Phys.orgās writeāup highlights that using synthetic datasets plus transfer learning reduces costly field labeling and improves crossāseason LAI monitoring accuracy. The study was published in Plant Phenomics on Feb 28, 2025.
Why it matters
LAI is a core indicator of crop growth and productivity, but models often break when applied across years or growth stages and typically need lots of labeled data. By fusing 3D physical simulation with deep learning and transfer learning, this work offers a more scalable way to monitor crop canopies over timeāpromising for precision agricultureāwhile the authors note future work is needed to extend beyond maize and to larger (e.g., satellite) scales.
ELI5 Summary
Imagine trying to tell how leafy a corn field is just from drone photos. āLeaf area index (LAI)ā is basically how thick a blanket of leaves covers the ground. The team first taught an AI inside a realistic computer-made field that simulates how light moves through 3D corn canopies. Then they gave it only a small set of real measurements from 2021 plus the drone images. Even so, in 2022 and 2023 the AI estimated leafiness well and beat older physics models and common ML methods. It only struggled a bit when the crop was super dense, sometimes reading LAI too low.
Did you find this news intellectually stimulating? |
Some affiliate links we endorse:
Stay curious,
Anthony Ao
The PhDLevel Team
āļøš» Powered by caffeine & curiosity
Hereās What a 1-Day Gutter Upgrade Should Cost You
Today, seniors (even on a fixed income or pension) can afford a modern gutter guard, along with 1-day installation and a 100% no-clog guarantee.
With this new website, you can āskipā the middleman, design a gutter guard thatās right for you, and get (fair) local pricing.