Meta-Reinforced Synthetic Data for One-Shot Fine-Grained Visual Recognition
arXiv:1911.07164
Abstract
One-shot fine-grained visual recognition often suffers from the problem of training data scarcity for new fine-grained classes. To alleviate this problem, an off-the-shelf image generator can be applied to synthesize additional training images, but these synthesized images are often not helpful for actually improving the accuracy of one-shot fine-grained recognition. This paper proposes a meta-learning framework to combine generated images with original images, so that the resulting ``hybrid'' training images can improve one-shot learning. Specifically, the generic image generator is updated by a few training instances of novel classes, and a Meta Image Reinforcing Network (MetaIRNet) is proposed to conduct one-shot fine-grained recognition as well as image reinforcement. The model is trained in an end-to-end manner, and our experiments demonstrate consistent improvement over baselines on one-shot fine-grained image classification benchmarks.
arXiv commentAccepted by Conference on Neural Information Processing System 2019
Cited by in corpus (6)
- TOAN: Target-Oriented Alignment Network for Fine-Grained Image Categorization with Few Labeled Samples
- Fine-Grained Image Analysis with Deep Learning: A Survey
- Bowtie Networks: Generative Modeling for Joint Few-Shot Recognition and Novel-View Synthesis
- Object-aware Long-short-range Spatial Alignment for Few-Shot Fine-Grained Image Classification
- DeltaGAN: Towards Diverse Few-shot Image Generation with Sample-Specific Delta
- Variational Feature Disentangling for Fine-Grained Few-Shot Classification