In this paper we evaluate the performance of scene detection techniques, starting from the classic precision/recall approach, moving to the better designed coverage/overflow measures, and finally proposing an improved metric, in order to solve frequently observed cases in which the numeric interpretation is different from the expected results. Numerical evaluation is performed on two recent proposals for automatic scene detection, and comparing them with a simple but effective novel approach. Experimental results are conducted to show how different measures may lead to different interpretations.
Measuring scene detection performance / Baraldi, Lorenzo; Grana, Costantino; Cucchiara, Rita. - 9117:(2015), pp. 395-403. (Intervento presentato al convegno 7th Iberian Conference on Pattern Recognition and Image Analysis tenutosi a Santiago de Compostela, Spain nel 17-19 June 2015) [10.1007/978-3-319-19390-8_45].
Measuring scene detection performance
BARALDI, LORENZO;GRANA, Costantino;CUCCHIARA, Rita
2015
Abstract
In this paper we evaluate the performance of scene detection techniques, starting from the classic precision/recall approach, moving to the better designed coverage/overflow measures, and finally proposing an improved metric, in order to solve frequently observed cases in which the numeric interpretation is different from the expected results. Numerical evaluation is performed on two recent proposals for automatic scene detection, and comparing them with a simple but effective novel approach. Experimental results are conducted to show how different measures may lead to different interpretations.File | Dimensione | Formato | |
---|---|---|---|
IbPRIA2015_cameraready.pdf
Open access
Tipologia:
Versione originale dell'autore proposta per la pubblicazione
Dimensione
934.67 kB
Formato
Adobe PDF
|
934.67 kB | Adobe PDF | Visualizza/Apri |
Pubblicazioni consigliate
I metadati presenti in IRIS UNIMORE sono rilasciati con licenza Creative Commons CC0 1.0 Universal, mentre i file delle pubblicazioni sono rilasciati con licenza Attribuzione 4.0 Internazionale (CC BY 4.0), salvo diversa indicazione.
In caso di violazione di copyright, contattare Supporto Iris