Assorted, Archetypal and Annotated Two Million (3A2M) Cooking Recipes Dataset based on Active Learning.
Nazmus SakibG. M. ShahariarMohsinul KabirMd. Kamrul HasanHasan MahmudPublished in: CoRR (2023)
Keyphrases
- active learning
- ground truth labels
- semi supervised
- data sets
- benchmark datasets
- training examples
- database
- learning algorithm
- pool based active learning
- million images
- annotated images
- sample selection
- selective sampling
- manually annotated
- experimental design
- synthetic datasets
- unlabeled data
- semi supervised learning
- supervised learning
- machine learning
- digital libraries
- high quality
- human actions
- feature set
- training dataset
- tens of thousands
- genetic algorithm
- feature space
- object recognition
- training data