Object-Level Representation Learning for Few-Shot Image Classification
This work addresses the problem of limited labeled data for image classification, offering a non-parametric method that improves accuracy without fine-tuning, though it is incremental in leveraging object-level relations.
The paper tackles few-shot image classification by leveraging object-level representations from an additional dataset to infer similarities for unseen categories, achieving absolute improvements of 8.5% and 2.7% on MiniImagenet for 5-way 1-shot and 5-way 5-shot tasks, respectively.
Few-shot learning that trains image classifiers over few labeled examples per category is a challenging task. In this paper, we propose to exploit an additional big dataset with different categories to improve the accuracy of few-shot learning over our target dataset. Our approach is based on the observation that images can be decomposed into objects, which may appear in images from both the additional dataset and our target dataset. We use the object-level relation learned from the additional dataset to infer the similarity of images in our target dataset with unseen categories. Nearest neighbor search is applied to do image classification, which is a non-parametric model and thus does not need fine-tuning. We evaluate our algorithm on two popular datasets, namely Omniglot and MiniImagenet. We obtain 8.5\% and 2.7\% absolute improvements for 5-way 1-shot and 5-way 5-shot experiments on MiniImagenet, respectively. Source code will be published upon acceptance.