Deep learning is driven by big data, which brings new opportunities for target classification, detection, semantic segmentation, instance segmentation, and regression in ecological resource research. However, there are still great challenges in the standardization and sharing of data, the universality and interpretability of algorithms, and the enrichment and intelligence of applications. A recent study clarified the aforementioned frontier issues, which was published in the ‘Science China Terrae’ journal.
- Why we’re so bad at daydreaming, and how to fix it
- Team of bioethicists and scientists suggests revisiting 14-day limit on human embryo
- COVID-19 lockdown linked to uptick in tobacco use
- An unstable working life affects the future mental health of young people
- Twistoptics–A new way to control optical nonlinearity