Matches in SemOpenAlex for { <https://semopenalex.org/work/W4283734354> ?p ?o ?g. }
- W4283734354 endingPage "102890" @default.
- W4283734354 startingPage "102890" @default.
- W4283734354 abstract "Mangroves are highly productive wetland ecosystems, located at the interlocking area of tropical and subtropical coastal zones. Accurately mapping the distribution, quality and quantity of species are crucial for mangrove management, protection, and restoration. This study proposed a mangrove species mapping approach by combining recursive feature elimination (RFE) with deep learning (DL) algorithms, and further assess the effectiveness of feature selection for DL (DeeplabV3+ and PSPNet) algorithm to improve classification accuracy under the high dimensional UAV image datasets. We constructed an ensemble learning models (SEL) by stacking five base models (Random Forest, XGBoost, LightGBM, CatBoost, and AdaBoost), and evaluate the classification ability of mangrove species between SEL and RFE-DL algorithms. Comparison of the classifications of mangrove species was to evaluate the accuracy differences between SEL and base models. Results indicated that: (1) RFE algorithm could improve the classification accuracy of DL algorithms. RFE-DL models using the optimal features achieved 94.8% of overall accuracy (OA), which was 0.2%-8.5% higher than only using the original multispectral bands; (2) SEL algorithm produced better classification performance than RFE-DL with a higher 1.6%-12.7% of overall accuracy. Mcnemar's test showed the classifications of mangrove species were significant differences between the three algorithms; (3) the SEL algorithm had a strong and stable ability for classifying mangrove species. The OA of six classification scenarios was from 75.5% to 92.2%, and the highest OA using SEL algorithm was 0.8%-4.2% higher than base models; (4) XGBoost algorithm had the highest importance, while AdaBoost had the lowest importance in the SEL-based classifications." @default.
- W4283734354 created "2022-07-01" @default.
- W4283734354 creator A5006581779 @default.
- W4283734354 creator A5025194582 @default.
- W4283734354 creator A5025522659 @default.
- W4283734354 creator A5042729829 @default.
- W4283734354 creator A5053344101 @default.
- W4283734354 creator A5058041620 @default.
- W4283734354 creator A5059723326 @default.
- W4283734354 creator A5074166685 @default.
- W4283734354 creator A5084848491 @default.
- W4283734354 date "2022-08-01" @default.
- W4283734354 modified "2023-09-29" @default.
- W4283734354 title "Comparison of RFE-DL and stacking ensemble learning algorithms for classifying mangrove species on UAV multispectral images" @default.
- W4283734354 cites W1442930683 @default.
- W4283734354 cites W1602811519 @default.
- W4283734354 cites W1964262728 @default.
- W4283734354 cites W1973897814 @default.
- W4283734354 cites W1980547211 @default.
- W4283734354 cites W2008469107 @default.
- W4283734354 cites W2071902553 @default.
- W4283734354 cites W2083615851 @default.
- W4283734354 cites W2135850590 @default.
- W4283734354 cites W2168846525 @default.
- W4283734354 cites W2591970477 @default.
- W4283734354 cites W2758210752 @default.
- W4283734354 cites W2768136227 @default.
- W4283734354 cites W2784301945 @default.
- W4283734354 cites W2804043600 @default.
- W4283734354 cites W28412257 @default.
- W4283734354 cites W2903282641 @default.
- W4283734354 cites W2935317424 @default.
- W4283734354 cites W2949747980 @default.
- W4283734354 cites W2950123062 @default.
- W4283734354 cites W2953297071 @default.
- W4283734354 cites W2955034228 @default.
- W4283734354 cites W2967896173 @default.
- W4283734354 cites W2971569792 @default.
- W4283734354 cites W2973118224 @default.
- W4283734354 cites W2979631444 @default.
- W4283734354 cites W2982275932 @default.
- W4283734354 cites W2993832485 @default.
- W4283734354 cites W3000457862 @default.
- W4283734354 cites W3016345523 @default.
- W4283734354 cites W3025406830 @default.
- W4283734354 cites W3041234790 @default.
- W4283734354 cites W3049157029 @default.
- W4283734354 cites W3100344990 @default.
- W4283734354 cites W3124539583 @default.
- W4283734354 cites W3139003984 @default.
- W4283734354 cites W3156668449 @default.
- W4283734354 cites W3190807025 @default.
- W4283734354 cites W3196547638 @default.
- W4283734354 cites W3207793649 @default.
- W4283734354 cites W4200429143 @default.
- W4283734354 cites W998093192 @default.
- W4283734354 doi "https://doi.org/10.1016/j.jag.2022.102890" @default.
- W4283734354 hasPublicationYear "2022" @default.
- W4283734354 type Work @default.
- W4283734354 citedByCount "4" @default.
- W4283734354 countsByYear W42837343542022 @default.
- W4283734354 countsByYear W42837343542023 @default.
- W4283734354 crossrefType "journal-article" @default.
- W4283734354 hasAuthorship W4283734354A5006581779 @default.
- W4283734354 hasAuthorship W4283734354A5025194582 @default.
- W4283734354 hasAuthorship W4283734354A5025522659 @default.
- W4283734354 hasAuthorship W4283734354A5042729829 @default.
- W4283734354 hasAuthorship W4283734354A5053344101 @default.
- W4283734354 hasAuthorship W4283734354A5058041620 @default.
- W4283734354 hasAuthorship W4283734354A5059723326 @default.
- W4283734354 hasAuthorship W4283734354A5074166685 @default.
- W4283734354 hasAuthorship W4283734354A5084848491 @default.
- W4283734354 hasBestOaLocation W42837343541 @default.
- W4283734354 hasConcept C110083411 @default.
- W4283734354 hasConcept C11413529 @default.
- W4283734354 hasConcept C119857082 @default.
- W4283734354 hasConcept C138885662 @default.
- W4283734354 hasConcept C153180895 @default.
- W4283734354 hasConcept C154945302 @default.
- W4283734354 hasConcept C169258074 @default.
- W4283734354 hasConcept C173163844 @default.
- W4283734354 hasConcept C18903297 @default.
- W4283734354 hasConcept C205649164 @default.
- W4283734354 hasConcept C2776401178 @default.
- W4283734354 hasConcept C2994302886 @default.
- W4283734354 hasConcept C33923547 @default.
- W4283734354 hasConcept C41008148 @default.
- W4283734354 hasConcept C41895202 @default.
- W4283734354 hasConcept C58640448 @default.
- W4283734354 hasConcept C68874143 @default.
- W4283734354 hasConcept C86803240 @default.
- W4283734354 hasConceptScore W4283734354C110083411 @default.
- W4283734354 hasConceptScore W4283734354C11413529 @default.
- W4283734354 hasConceptScore W4283734354C119857082 @default.
- W4283734354 hasConceptScore W4283734354C138885662 @default.
- W4283734354 hasConceptScore W4283734354C153180895 @default.
- W4283734354 hasConceptScore W4283734354C154945302 @default.
- W4283734354 hasConceptScore W4283734354C169258074 @default.