Centre de documentation
scientifique

Deep traffic light detection by overlaying synthetic context on arbitrary natural images / Jean Pablo Vieira de Mello in Computers and graphics, vol 94 n° 1 (February 2021)  

Public [article]  inComputers and graphics > vol 94 n° 1 (February 2021) . - pp 76 - 86 Titre : Deep traffic light detection by overlaying synthetic context on arbitrary natural images Type de document : Article/Communication Auteurs : Jean Pablo Vieira de Mello, Auteur ; Lucas Tabelini, Auteur ; Rodrigo F. Berriel, Auteur Année de publication : 2021 Article en page(s) : pp 76 - 86 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique [Termes IGN] analyse d'image orientée objet [Termes IGN] apprentissage profond [Termes IGN] détection d'objet [Termes IGN] échantillonnage d'image [Termes IGN] feu de circulation [Termes IGN] image à haute résolution [Termes IGN] navigation autonome [Termes IGN] signalisation routière [Termes IGN] trafic routier Résumé : (auteur) Deep neural networks come as an effective solution to many problems associated with autonomous driving. By providing real image samples with traffic context to the network, the model learns to detect and classify elements of interest, such as pedestrians, traffic signs, and traffic lights. However, acquiring and annotating real data can be extremely costly in terms of time and effort. In this context, we propose a method to generate artificial traffic-related training data for deep traffic light detectors. This data is generated using basic non-realistic computer graphics to blend fake traffic scenes on top of arbitrary image backgrounds that are not related to the traffic domain. Thus, a large amount of training data can be generated without annotation efforts. Furthermore, it also tackles the intrinsic data imbalance problem in traffic light datasets, caused mainly by the low amount of samples of the yellow state. Experiments show that it is possible to achieve results comparable to those obtained with real training data from the problem domain, yielding an average mAP and an average F1-score which are each nearly 4 p.p. higher than the respective metrics obtained with a real-world reference model. Numéro de notice : A2021-151 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Article DOI : 10.1016/j.cag.2020.09.012 Date de publication en ligne : 09/10/2020 En ligne : https://doi.org/10.1016/j.cag.2020.09.012 Format de la ressource électronique : url article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=97027 [article]