Combating noisy labels in object detection datasets.
Krystian ChachulaAdam PopowiczJakub LyskawaBartlomiej OlberPiotr FratczakKrystian RadlakPublished in: CoRR (2022)
Keyphrases
- object detection
- pascal voc
- training data
- scene recognition
- benchmark datasets
- object categories
- face detection
- object class
- data sets
- data mining
- weakly labeled
- label noise
- scene understanding
- object classification
- pedestrian detection
- multi class
- object recognition
- computer vision
- object classes
- missing data
- synthetic datasets
- support vector machine
- training set
- noise free
- learning algorithm
- images with ground truth