Gestalt-Guided Image Understanding for Few-Shot Learning

Due to the scarcity of available data, deep learning does not perform well on few-shot tasks. However, human can quickly learn feature a new category from very few samples. Nevertheless, previous work has rarely considered how mimic cognitive behavior and apply it learning. This paper introduces Gestalt psychology proposes Gestalt-Guided Image Understanding, plug-and-play method called GGIU. Referring principle totality law closure in psychology, we design Totality-Guided Understanding Closure-Guided extract image features. After that, estimation module is used estimate accurate features images. Extensive experiments demonstrate that our improve performance existing models effectively flexibly without retraining or fine-tuning. Our code released https://github.com/skingorz/GGIU .