Improved few-shot aerial image classification using rotation-aware component pooling
Component Pooling Trained Across Orientations for Few-Shot Remote Sensing Scene Classification
Computer Vision and Pattern Recognition
Summary
Recognizing new types of aerial scenes from just a few labelled pictures is hard, partly because images don’t always have the same orientation. The authors created a method called RoCoP that groups parts of the image into meaningful components and learns to classify images by looking at many rotated versions. This approach better focuses on important objects and ignores irrelevant background, boosting accuracy without extra computing costs during use. Tests on several aerial image datasets showed RoCoP works better than other few-shot learning methods, except when very little base data is available.
What this means in practice
- •For remote sensing analysts: Enhance classification of land-use scenes when only a few labeled satellite images per category are available, using rotation-aware feature grouping.
- •For environmental monitoring teams: Improve detection of novel landscape features from limited aerial image data despite arbitrary image rotations.
Authors
Bingchen Yan, Wenheng Guo, Yuye Zhou
Abstract
Few-shot remote sensing scene classification (FSRSSC) aims to recognize novel scene categories from only a few labeled images. Most FSRSSC methods summarize the feature map of an image by global average pooling, which treats the few objects that define an aerial scene and the large background around them alike, and they rarely consider that aerial images have no canonical orientation. To address these issues, we propose RoCoP, a parameter-free component pooling network trained across orientations. RoCoP consists of a feature extractor, a non-negative decomposition module (NDM) and a component scoring module (CSM). NDM factorizes the local descriptors of each image into a few additive components with a deterministic initialization and a single multiplicative update, so that the positions of one object are grouped together. CSM weights each component by its activation magnitude and converts the weights into a soft mask for pooling. Because both modules are invariant to permutations of feature positions, the only rotation-sensitive part of the pipeline is the backbone, which we train to classify randomly rotated views of every image. Experiments on NWPU-RESISC45, AID and UC Merced show that component pooling and rotated-view training are complementary: together they improve a strong transfer-learning baseline by about 2.5 points in 5-way 1-shot accuracy on NWPU-RESISC45 and AID, without adding parameters at inference. Under an identical protocol, RoCoP outperforms re-implemented metric-based and task-adaptive methods, and NDM and CSM also improve existing methods when plugged into them. We further identify the regime in which the method does not apply, namely very small base sets.