ORBIT: A Real-World Few-Shot Dataset for Teachable Object Recognition.
Daniela MassicetiLuisa M. ZintgrafJohn BronskillLida TheodorouMatthew Tobias HarrisEdward CutrellCecily MorrisonKatja HofmannSimone StumpfPublished in: CoRR (2021)
Keyphrases
- object recognition
- real world
- synthetic datasets
- wide range
- computer vision
- pascal voc
- video sequences
- case study
- low level features
- data sets
- natural images
- d objects
- synthetic data
- benchmark datasets
- data mining
- training dataset
- visual features
- event detection
- template matching
- real life
- object categories
- feature detection
- video shots
- model construction
- feature selection