7 ms·
We're not talking about putting the pictures into ChatGPT here. Anyway, from their tests using standard CNN practices: "Our best model achieved on the test set
by niklasrde 3y ago
We're not talking about putting the pictures into ChatGPT here.
Anyway, from their tests using standard CNN practices:
"Our best model achieved on the test set an F1 score of 0.97 (accuracy = 97.5%) for the classification task and a R2 score of 0.84 (RMSE = 21.9 m, or about 1 image pixel) for the length-estimation task."
- mistrial9 3y agono - enhanced resolution is very iffy and subject to gaming etc secondly, there is no perfect way to measure accuracy of remote sensing for many reasons
- addandsubtract 3y agoThey're tracking a vessel over time, though. Surely any errors and inaccuracies are negated over a long enough time horizon.