Active learning with MaskAL reduces annotation effort for training Mask R-CNN on a broccoli dataset with visually similar classes.
Pieter M. BlokGert KootstraHakim Elchaoui ElghorBoubacar DialloFrits K. van EvertEldert J. van HentenPublished in: Comput. Electron. Agric. (2022)
Keyphrases
- annotation effort
- active learning
- visually similar
- training data
- automatically extracted
- learning algorithm
- training examples
- machine learning
- supervised learning
- training set
- semi supervised
- relevance feedback
- learning process
- visual features
- labeled data
- unlabeled data
- image set
- computer vision
- image regions
- visual similarity
- image retrieval
- class labels
- face recognition