Semantic Segmentation Using Pixel-Wise Adaptive Label Smoothing via Self-Knowledge Distillation for Limited Labeling Data

To achieve high performance, most deep convolutional neural networks (DCNNs) require a significant amount of training data with ground truth labels. However, creating ground-truth labels for semantic segmentation requires more time, human effort, and cost compared with other tasks such as classifica...

Πλήρης περιγραφή

Λεπτομέρειες βιβλιογραφικής εγγραφής

Κύριοι συγγραφείς:	Sangyong Park, Jaeseon Kim, Yong Seok Heo
Μορφή:	Artigo
Γλώσσα:	Inglês
Έκδοση:	MDPI AG 2022-03-01
Σειρά:	Sensors
Θέματα:	semantic segmentation limited training data self-knowledge distillation regularization
Διαθέσιμο Online:	https://www.mdpi.com/1424-8220/22/7/2623
Ετικέτες:	Προσθήκη ετικέτας Δεν υπάρχουν, Καταχωρήστε ετικέτα πρώτοι!