Image Deformation Meta-Networks for One-Shot Learning

By Zitian Chen, Yanwei Fu, Yu-Xiong Wang, Lin Ma, Martial Hebert. 2019

252 citations recorded by the source corpus.

Humans can robustly learn novel visual concepts even when images undergo various deformations and loose certain information. Mimicking the same behavior and synthesizing deformed instances of new concepts may help visual recognition systems perform better one-shot learning, i.e., learning concepts from one or few examples. Our key insight is that, while the deformed images may not be visually realistic, they still maintain critical semantic information and contribute significantly to formulating classifier decision boundaries. Inspired by the recent progress of meta-learning, we combine a meta-learner with an image deformation sub-network that produces additional training examples, and optimize both models in an end-to-end manner. The deformation sub-network learns to deform images by fusing a pair of images --- a probe image that keeps the visual content and a gallery image that diversifies the deformations. We demonstrate results on the widely used one-shot learning benchmarks (miniImageNet and ImageNet 1K Challenge datasets), which significantly outperform state-of-the-art approaches.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Incremental Few-Shot Object Detection (2020, neutral): “Note that there are many more recent FSL methods [6, 26, 7, 17, 47, 55, 50, 27, 39, 11, 16] that produce better performance on image classification tasks than MAML.”
Interventional Few-Shot Learning (2020, methodology): “5) According to Table 1 and Table 2, it is clear that our k -NN+IFSL outperforms IdeMe-Net [14] using the same pre-trained ResNet-10.”
A Comprehensive Survey of Few-shot Learning: Evolution, Applications, Challenges, and Opportunities (2022, background): “[15] proposes an end-to-end approach to partitioning images as a whole inspired with the idea of MIXUP [54], which argues that images preserve important semantic information even after they have undergone various distortions.”
Bowtie Networks: Generative Modeling for Joint Few-Shot Recognition and Novel-View Synthesis (2020, background): “Many algorithms have been proposed to solve this problem [6, 36, 29], including the recent efforts on leveraging generative models [37, 45, 14, 33, 4].”
Learning What Not to Segment: A New Perspective on Few-Shot Segmentation (2022, methodology): “On this basis, FSL approaches can be further subdivided into three branches: (i) metric-based [24, 49, 50], (ii) optimizationbased [11, 19, 42], and (iii) augmentation-based [5, 6].”
Leveraging the Feature Distribution in Transfer-based Few-Shot Learning (2020, methodology): “Other works [36, 4] utilize data augmentation techniques to artificially increase the size of the training datasets.”
Meta-Learning to Detect Rare Objects (2019, methodology): “Our approach falls under the umbrella of meta-learning [62, 63, 51, 52, 1, 50, 67, 70, 4, 42, 17, 58, 71, 22, 40, 34, 60, 28, 68, 53, 48, 8, 33], and is most related to the work on model parameter estimation [70].”
Long-tailed Recognition by Routing Diverse Distribution-Aware Experts (2020, unknown): “A BSTRACT (b) Histograms of the largest softmax probability of the other classes (the hardest negative) per instance.”